In [30]:
import os
import numpy as np
import pandas as pd
import multiprocessing
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.cluster import KMeans
from scipy.spatial.distance import pdist, squareform
from scipy.stats import spearmanr
from sklearn.metrics.pairwise import cosine_similarity
import h5py
# Define directories
#"\\wsl.localhost\Ubuntu\home\abbasali\synthetic data\generated_data"
input_dir = "/home/abbasali/synthetic data/generated_data_llama" # WSL path
output_dir = "/home/abbasali/synthetic data/embeddings_llama" # WSL path
# Ensure directories exist
if not os.path.exists(input_dir):
raise FileNotFoundError(f"❌ Directory not found: {input_dir}")
if not os.path.exists(output_dir):
raise FileNotFoundError(f"❌ Directory not found: {output_dir}")
print("✅ Directories verified.")
# Load all CSV datasets
dataframes = {}
csv_files = [f for f in os.listdir(input_dir) if f.endswith(".csv")]
for file in csv_files:
file_path = os.path.join(input_dir, file)
dataset_name = os.path.splitext(file)[0] # Extract dataset name (without .csv)
dataframes[dataset_name] = pd.read_csv(file_path)
print(f"✅ Loaded {len(dataframes)} CSV files.")
# Load all embeddings from HDF5 files
embeddings_data = {}
embedding_files = [f for f in os.listdir(output_dir) if f.endswith(".h5")]
for file in embedding_files:
file_parts = file.split("_")
dataset_name = "_".join(file_parts[1:]).replace(".h5", "")
if dataset_name in dataframes: # Only keep embeddings that match available datasets
if dataset_name not in embeddings_data:
embeddings_data[dataset_name] = {}
with h5py.File(os.path.join(output_dir, file), 'r') as h5f:
embeddings = h5f['embeddings'][:]
embeddings_data[dataset_name]['llama'] = embeddings # Assuming 'llama' model
print(f"✅ Loaded {len(embeddings_data)} embedding sets.")
def find_optimal_clusters(data, max_k=10):
"""
Uses the Elbow Method and Silhouette Score to find the optimal number of clusters.
Parameters:
- data: numpy array (data to cluster)
- max_k: int (maximum number of clusters to test)
Returns:
- optimal_k: int (optimal number of clusters)
"""
distortions = []
silhouette_scores = []
K_range = range(2, min(max_k, len(data))) # Ensure K is valid for the data size
for k in K_range:
kmeans = KMeans(n_clusters=k, random_state=42, n_init=10).fit(data)
distortions.append(kmeans.inertia_)
if len(data) > k: # Avoid errors for small data
silhouette_avg = silhouette_score(data, kmeans.labels_)
silhouette_scores.append(silhouette_avg)
# Find elbow point using second derivative method
if len(distortions) > 2:
elbow_point = np.diff(distortions, 2).argmin() + 2
else:
elbow_point = 2 # Default to 2 clusters if we lack enough points
# Choose best number of clusters
if silhouette_scores:
best_silhouette = K_range[np.argmax(silhouette_scores)]
optimal_k = best_silhouette if best_silhouette > elbow_point else elbow_point
else:
optimal_k = elbow_point
return optimal_k
def compute_cluster_similarity(dataset_name):
"""Computes cluster relationships between original data and embeddings using a fixed number of clusters."""
results = []
# Load original dataset
df = dataframes[dataset_name]
# Select continuous numerical features
continuous_features = df.iloc[:, :3].values
# Limit sample size
sample_size = min(300, continuous_features.shape[0])
continuous_sample = continuous_features[:sample_size]
# Define a fixed number of clusters
fixed_num_clusters = 5 # Adjust based on your data characteristics
# Apply KMeans clustering to the original data
kmeans_orig = KMeans(n_clusters=fixed_num_clusters, random_state=42, n_init=10).fit(continuous_sample)
centroids_orig = kmeans_orig.cluster_centers_
# Compute pairwise distances between original centroids
dist_orig = squareform(pdist(centroids_orig, metric='euclidean'))
# Process each model's embeddings
for model_name, embeddings in embeddings_data[dataset_name].items():
print(f"📊 Processing: {dataset_name} (Model: {model_name})")
# Reduce sample size for embeddings
embeddings_sample = embeddings[:sample_size]
# Apply KMeans clustering to the embeddings
kmeans_emb = KMeans(n_clusters=fixed_num_clusters, random_state=42, n_init=10).fit(embeddings_sample)
centroids_emb = kmeans_emb.cluster_centers_
# Compute pairwise distances between embedding centroids
dist_emb = squareform(pdist(centroids_emb, metric='euclidean'))
# Compute Spearman correlation
correlation, _ = spearmanr(dist_orig.flatten(), dist_emb.flatten())
# Compute cosine similarity
cosine_sim = cosine_similarity(dist_orig.flatten().reshape(1, -1), dist_emb.flatten().reshape(1, -1))[0][0]
# Store results
results.append({
"Dataset": dataset_name,
"Model": model_name,
"Fixed Clusters": fixed_num_clusters,
"Distance Correlation": round(correlation, 3),
"Cosine Similarity": round(cosine_sim, 3)
})
return results
if __name__ == "__main__":
num_workers = min(multiprocessing.cpu_count(), len(dataframes)) # Limit to CPU count
with multiprocessing.Pool(num_workers) as pool:
all_results = pool.map(compute_cluster_similarity, list(dataframes.keys()))
# Flatten list of lists
final_results = [item for sublist in all_results for item in sublist]
# Convert to DataFrame
results_df = pd.DataFrame(final_results)
print("✅ Cluster similarity analysis completed.")
# Summary statistics by model
summary = results_df.groupby("Model")[["Distance Correlation", "Cosine Similarity"]].agg(["mean", "std"]).reset_index()
# Save results
results_df.to_csv("cluster_similarity_results.csv", index=False)
summary.to_csv("model_summary_results.csv", index=False)
print("✅ Results saved to CSV files.")
# Set visualization style
sns.set(style="whitegrid")
# Boxplot for Distance Correlation across Models
plt.figure(figsize=(12, 6))
sns.boxplot(x="Model", y="Distance Correlation", data=results_df, palette="viridis")
plt.xticks(rotation=45, fontsize=14)
plt.yticks(fontsize=14)
plt.xlabel("Model", fontsize=16)
plt.ylabel("Distance Correlation", fontsize=16)
plt.title("Distance Correlation Across Models", fontsize=18)
plt.show()
# Boxplot for Cosine Similarity across Models
plt.figure(figsize=(12, 6))
sns.boxplot(x="Model", y="Cosine Similarity", data=results_df, palette="coolwarm")
plt.xticks(rotation=45, fontsize=14)
plt.yticks(fontsize=14)
plt.xlabel("Model", fontsize=16)
plt.ylabel("Cosine Similarity", fontsize=16)
plt.title("Cosine Similarity Across Models", fontsize=18)
plt.show()
✅ Directories verified. ✅ Loaded 14 CSV files. ✅ Loaded 14 embedding sets. 📊 Processing: data_12_exponential (Model: llama) 📊 Processing: data_53_quadratic (Model: llama) 📊 Processing: data_75_exponential (Model: llama) 📊 Processing: data_50_exponential (Model: llama) 📊 Processing: data_85_quadratic (Model: llama) 📊 Processing: data_24_exponential (Model: llama) 📊 Processing: data_51_quadratic (Model: llama) 📊 Processing: data_58_quadratic (Model: llama) 📊 Processing: data_49_linear (Model: llama)📊 Processing: data_70_linear (Model: llama) 📊 Processing: data_15_quadratic (Model: llama) 📊 Processing: data_47_cubic (Model: llama) 📊 Processing: data_91_linear (Model: llama) 📊 Processing: data_80_cubic (Model: llama) ✅ Cluster similarity analysis completed. ✅ Results saved to CSV files.
/tmp/ipykernel_973/3571337628.py:176: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(x="Model", y="Distance Correlation", data=results_df, palette="viridis")
/tmp/ipykernel_973/3571337628.py:186: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(x="Model", y="Cosine Similarity", data=results_df, palette="coolwarm")
In [20]:
import os
import numpy as np
import pandas as pd
import h5py
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE
from sklearn.impute import SimpleImputer
from scipy.spatial.distance import pdist, squareform
from scipy.stats import spearmanr
from sklearn.metrics.pairwise import cosine_similarity
# ✅ Ask user explicitly to re-run clustering or llama processing
def get_yes_no_input(prompt):
while True:
user_input = input(prompt).strip().lower()
if user_input in ["yes", "y"]:
return True
elif user_input in ["no", "n"]:
return False
print("⚠️ Invalid input! Please enter 'yes' or 'no'.")
RE_RUN = get_yes_no_input("🔄 Do you want to re-run clustering analysis? (yes/no): ")
RE_RUN_LLAMA = get_yes_no_input("🔄 Do you want to re-run Llama processing? (yes/no): ")
# ✅ Define dataset-embedding pairs with correct file extensions
processing_sets = [
{
"input_dir": "/home/abbasali/synthetic data/generated_data",
"output_dir": "/home/abbasali/synthetic data/embeddings",
"file_ext": ".npy",
"models": ["bert", "ernie", "gatortron", "roberta", "t5"]
},
{
"input_dir": "/home/abbasali/synthetic data/generated_data_llama",
"output_dir": "/home/abbasali/synthetic data/embeddings_llama",
"file_ext": ".h5",
"models": ["llama"]
}
]
# ✅ Load previous results if available
RESULTS_FILE = "cluster_similarity_results1.csv"
if os.path.exists(RESULTS_FILE) and not RE_RUN:
existing_results = pd.read_csv(RESULTS_FILE)
processed_files = set(zip(existing_results["Dataset"], existing_results["Model"]))
else:
existing_results = pd.DataFrame()
processed_files = set()
def load_data(input_dir, output_dir, file_ext, models):
"""Loads CSV datasets and dynamically finds embeddings in the output directory."""
print(f"📂 Processing {input_dir} -> {output_dir}")
if not os.path.exists(input_dir) or not os.path.exists(output_dir):
print(f"❌ ERROR: Missing directory - {input_dir} or {output_dir}")
return {}, {}
existing_files = set(os.listdir(output_dir))
csv_files = [f for f in os.listdir(input_dir) if f.endswith(".csv")]
if not csv_files:
print(f"⚠️ No CSV datasets found in {input_dir}")
dataframes = {os.path.splitext(f)[0]: pd.read_csv(os.path.join(input_dir, f)) for f in csv_files}
print(f"✅ Loaded {len(dataframes)} CSV datasets from {input_dir}.")
embeddings_data = {}
for filename in csv_files:
dataset_name = os.path.splitext(filename)[0]
for model_name in models:
# ✅ Dynamically find matching filenames in the directory
if model_name == "llama":
matched_files = [f for f in existing_files if f"embeddings_data_{dataset_name}" in f and f.endswith(".h5")]
else:
matched_files = [f for f in existing_files if f.startswith(f"embeddings_{model_name}_{dataset_name}") and f.endswith(".npy")]
if not matched_files:
print(f"⚠️ WARNING: Missing embeddings for {dataset_name} (Model: {model_name})")
continue
embeddings_filepath = os.path.join(output_dir, matched_files[0])
if (dataset_name, model_name) in processed_files:
print(f"⚠️ Skipping already processed file: {dataset_name} (Model: {model_name})")
continue
try:
if model_name == "llama":
with h5py.File(embeddings_filepath, 'r') as h5f:
embeddings = h5f['embeddings'][:]
else:
embeddings = np.load(embeddings_filepath)
if embeddings.size == 0:
print(f"❌ ERROR: Empty embeddings file detected: {embeddings_filepath}")
continue
# ✅ Ensure embeddings are 2D
if embeddings.ndim > 2:
embeddings = embeddings.reshape(embeddings.shape[0], -1)
if dataset_name not in embeddings_data:
embeddings_data[dataset_name] = {}
embeddings_data[dataset_name][model_name] = embeddings
print(f"✅ Loaded embeddings for {dataset_name} (Model: {model_name})")
except Exception as e:
print(f"❌ ERROR: Could not load {embeddings_filepath} - {e}")
return dataframes, embeddings_data
def compute_cluster_similarity(dataframes, embeddings_data):
"""Computes cluster relationships using PCA, t-SNE, and KMeans clustering."""
results = []
for dataset_name, df in dataframes.items():
if df.empty:
print(f"⚠️ Skipping empty dataset: {dataset_name}")
continue
continuous_features = df.iloc[:, :3].values
sample_size = min(300, continuous_features.shape[0])
continuous_sample = SimpleImputer(strategy="mean").fit_transform(continuous_features[:sample_size])
optimal_clusters = min(40, len(np.unique(continuous_sample, axis=0)))
optimal_clusters = max(2, optimal_clusters)
kmeans_orig = KMeans(n_clusters=optimal_clusters, random_state=42, n_init=10).fit(continuous_sample)
dist_orig = squareform(pdist(kmeans_orig.cluster_centers_, metric='euclidean'))
for model_name, embeddings in embeddings_data.get(dataset_name, {}).items():
print(f"📊 Processing: {dataset_name} (Model: {model_name})")
embeddings_sample = SimpleImputer(strategy="mean").fit_transform(embeddings[:sample_size])
pca_embeddings = PCA(n_components=50).fit_transform(embeddings_sample)
tsne_embeddings = TSNE(n_components=2, perplexity=30, random_state=42).fit_transform(pca_embeddings)
adjusted_clusters = min(optimal_clusters, len(tsne_embeddings), 10)
adjusted_clusters = max(2, adjusted_clusters)
kmeans_emb = KMeans(n_clusters=adjusted_clusters, random_state=42, n_init=10).fit(tsne_embeddings)
dist_emb = squareform(pdist(kmeans_emb.cluster_centers_, metric='euclidean'))
min_size = min(dist_orig.size, dist_emb.size)
dist_orig_resized = dist_orig.flatten()[:min_size]
dist_emb_resized = dist_emb.flatten()[:min_size]
correlation, _ = spearmanr(dist_orig_resized, dist_emb_resized)
cosine_sim = cosine_similarity(dist_orig_resized.reshape(1, -1), dist_emb_resized.reshape(1, -1))[0][0]
results.append({
"Dataset": dataset_name,
"Function Type": dataset_name.split("_")[-1],
"Model": model_name,
"Optimal Clusters": optimal_clusters,
"Adjusted Clusters (Embeddings)": adjusted_clusters,
"Distance Correlation": round(correlation, 3),
"Cosine Similarity": round(cosine_sim, 3)
})
return results
# ✅ Run pipeline
all_results = []
for set_info in processing_sets:
dataframes, embeddings_data = load_data(set_info["input_dir"], set_info["output_dir"], set_info["file_ext"], set_info["models"])
all_results.extend(compute_cluster_similarity(dataframes, embeddings_data))
pd.DataFrame(all_results).to_csv("cluster_similarity_results1.csv", index=False)
print("✅ Results saved to cluster_similarity_results1.csv")
📂 Processing /home/abbasali/synthetic data/generated_data -> /home/abbasali/synthetic data/embeddings ✅ Loaded 100 CSV datasets from /home/abbasali/synthetic data/generated_data. ✅ Loaded embeddings for data_79_quadratic (Model: bert) ✅ Loaded embeddings for data_79_quadratic (Model: ernie) ✅ Loaded embeddings for data_79_quadratic (Model: gatortron) ✅ Loaded embeddings for data_79_quadratic (Model: roberta) ✅ Loaded embeddings for data_79_quadratic (Model: t5) ✅ Loaded embeddings for data_91_linear (Model: bert) ✅ Loaded embeddings for data_91_linear (Model: ernie) ✅ Loaded embeddings for data_91_linear (Model: gatortron) ✅ Loaded embeddings for data_91_linear (Model: roberta) ✅ Loaded embeddings for data_91_linear (Model: t5) ✅ Loaded embeddings for data_44_exponential (Model: bert) ✅ Loaded embeddings for data_44_exponential (Model: ernie) ✅ Loaded embeddings for data_44_exponential (Model: gatortron) ✅ Loaded embeddings for data_44_exponential (Model: roberta) ✅ Loaded embeddings for data_44_exponential (Model: t5) ✅ Loaded embeddings for data_42_linear (Model: bert) ✅ Loaded embeddings for data_42_linear (Model: ernie) ✅ Loaded embeddings for data_42_linear (Model: gatortron) ✅ Loaded embeddings for data_42_linear (Model: roberta) ✅ Loaded embeddings for data_42_linear (Model: t5) ✅ Loaded embeddings for data_88_exponential (Model: bert) ✅ Loaded embeddings for data_88_exponential (Model: ernie) ✅ Loaded embeddings for data_88_exponential (Model: gatortron) ✅ Loaded embeddings for data_88_exponential (Model: roberta) ✅ Loaded embeddings for data_88_exponential (Model: t5) ✅ Loaded embeddings for data_43_linear (Model: bert) ✅ Loaded embeddings for data_43_linear (Model: ernie) ✅ Loaded embeddings for data_43_linear (Model: gatortron) ✅ Loaded embeddings for data_43_linear (Model: roberta) ✅ Loaded embeddings for data_43_linear (Model: t5) ✅ Loaded embeddings for data_68_exponential (Model: bert) ✅ Loaded embeddings for data_68_exponential (Model: ernie) ✅ Loaded embeddings for data_68_exponential (Model: gatortron) ✅ Loaded embeddings for data_68_exponential (Model: roberta) ✅ Loaded embeddings for data_68_exponential (Model: t5) ✅ Loaded embeddings for data_48_cubic (Model: bert) ✅ Loaded embeddings for data_48_cubic (Model: ernie) ✅ Loaded embeddings for data_48_cubic (Model: gatortron) ✅ Loaded embeddings for data_48_cubic (Model: roberta) ✅ Loaded embeddings for data_48_cubic (Model: t5) ✅ Loaded embeddings for data_13_cubic (Model: bert) ✅ Loaded embeddings for data_13_cubic (Model: ernie) ✅ Loaded embeddings for data_13_cubic (Model: gatortron) ✅ Loaded embeddings for data_13_cubic (Model: roberta) ✅ Loaded embeddings for data_13_cubic (Model: t5) ✅ Loaded embeddings for data_33_quadratic (Model: bert) ✅ Loaded embeddings for data_33_quadratic (Model: ernie) ✅ Loaded embeddings for data_33_quadratic (Model: gatortron) ✅ Loaded embeddings for data_33_quadratic (Model: roberta) ✅ Loaded embeddings for data_33_quadratic (Model: t5) ✅ Loaded embeddings for data_84_cubic (Model: bert) ✅ Loaded embeddings for data_84_cubic (Model: ernie) ✅ Loaded embeddings for data_84_cubic (Model: gatortron) ✅ Loaded embeddings for data_84_cubic (Model: roberta) ✅ Loaded embeddings for data_84_cubic (Model: t5) ✅ Loaded embeddings for data_74_linear (Model: bert) ✅ Loaded embeddings for data_74_linear (Model: ernie) ✅ Loaded embeddings for data_74_linear (Model: gatortron) ✅ Loaded embeddings for data_74_linear (Model: roberta) ✅ Loaded embeddings for data_74_linear (Model: t5) ✅ Loaded embeddings for data_21_quadratic (Model: bert) ✅ Loaded embeddings for data_21_quadratic (Model: ernie) ✅ Loaded embeddings for data_21_quadratic (Model: gatortron) ✅ Loaded embeddings for data_21_quadratic (Model: roberta) ✅ Loaded embeddings for data_21_quadratic (Model: t5) ✅ Loaded embeddings for data_40_exponential (Model: bert) ✅ Loaded embeddings for data_40_exponential (Model: ernie) ✅ Loaded embeddings for data_40_exponential (Model: gatortron) ✅ Loaded embeddings for data_40_exponential (Model: roberta) ✅ Loaded embeddings for data_40_exponential (Model: t5) ✅ Loaded embeddings for data_5_cubic (Model: bert) ✅ Loaded embeddings for data_5_cubic (Model: ernie) ✅ Loaded embeddings for data_5_cubic (Model: gatortron) ✅ Loaded embeddings for data_5_cubic (Model: roberta) ✅ Loaded embeddings for data_5_cubic (Model: t5) ✅ Loaded embeddings for data_17_exponential (Model: bert) ✅ Loaded embeddings for data_17_exponential (Model: ernie) ✅ Loaded embeddings for data_17_exponential (Model: gatortron) ✅ Loaded embeddings for data_17_exponential (Model: roberta) ✅ Loaded embeddings for data_17_exponential (Model: t5) ✅ Loaded embeddings for data_0_exponential (Model: bert) ✅ Loaded embeddings for data_0_exponential (Model: ernie) ✅ Loaded embeddings for data_0_exponential (Model: gatortron) ✅ Loaded embeddings for data_0_exponential (Model: roberta) ✅ Loaded embeddings for data_0_exponential (Model: t5) ✅ Loaded embeddings for data_30_cubic (Model: bert) ✅ Loaded embeddings for data_30_cubic (Model: ernie) ✅ Loaded embeddings for data_30_cubic (Model: gatortron) ✅ Loaded embeddings for data_30_cubic (Model: roberta) ✅ Loaded embeddings for data_30_cubic (Model: t5) ✅ Loaded embeddings for data_34_cubic (Model: bert) ✅ Loaded embeddings for data_34_cubic (Model: ernie) ✅ Loaded embeddings for data_34_cubic (Model: gatortron) ✅ Loaded embeddings for data_34_cubic (Model: roberta) ✅ Loaded embeddings for data_34_cubic (Model: t5) ✅ Loaded embeddings for data_52_cubic (Model: bert) ✅ Loaded embeddings for data_52_cubic (Model: ernie) ✅ Loaded embeddings for data_52_cubic (Model: gatortron) ✅ Loaded embeddings for data_52_cubic (Model: roberta) ✅ Loaded embeddings for data_52_cubic (Model: t5) ✅ Loaded embeddings for data_16_cubic (Model: bert) ✅ Loaded embeddings for data_16_cubic (Model: ernie) ✅ Loaded embeddings for data_16_cubic (Model: gatortron) ✅ Loaded embeddings for data_16_cubic (Model: roberta) ✅ Loaded embeddings for data_16_cubic (Model: t5) ✅ Loaded embeddings for data_72_linear (Model: bert) ✅ Loaded embeddings for data_72_linear (Model: ernie) ✅ Loaded embeddings for data_72_linear (Model: gatortron) ✅ Loaded embeddings for data_72_linear (Model: roberta) ✅ Loaded embeddings for data_72_linear (Model: t5) ✅ Loaded embeddings for data_89_linear (Model: bert) ✅ Loaded embeddings for data_89_linear (Model: ernie) ✅ Loaded embeddings for data_89_linear (Model: gatortron) ✅ Loaded embeddings for data_89_linear (Model: roberta) ✅ Loaded embeddings for data_89_linear (Model: t5) ✅ Loaded embeddings for data_55_linear (Model: bert) ✅ Loaded embeddings for data_55_linear (Model: ernie) ✅ Loaded embeddings for data_55_linear (Model: gatortron) ✅ Loaded embeddings for data_55_linear (Model: roberta) ✅ Loaded embeddings for data_55_linear (Model: t5) ✅ Loaded embeddings for data_22_quadratic (Model: bert) ✅ Loaded embeddings for data_22_quadratic (Model: ernie) ✅ Loaded embeddings for data_22_quadratic (Model: gatortron) ✅ Loaded embeddings for data_22_quadratic (Model: roberta) ✅ Loaded embeddings for data_22_quadratic (Model: t5) ✅ Loaded embeddings for data_65_exponential (Model: bert) ✅ Loaded embeddings for data_65_exponential (Model: ernie) ✅ Loaded embeddings for data_65_exponential (Model: gatortron) ✅ Loaded embeddings for data_65_exponential (Model: roberta) ✅ Loaded embeddings for data_65_exponential (Model: t5) ✅ Loaded embeddings for data_14_linear (Model: bert) ✅ Loaded embeddings for data_14_linear (Model: ernie) ✅ Loaded embeddings for data_14_linear (Model: gatortron) ✅ Loaded embeddings for data_14_linear (Model: roberta) ✅ Loaded embeddings for data_14_linear (Model: t5) ✅ Loaded embeddings for data_25_exponential (Model: bert) ✅ Loaded embeddings for data_25_exponential (Model: ernie) ✅ Loaded embeddings for data_25_exponential (Model: gatortron) ✅ Loaded embeddings for data_25_exponential (Model: roberta) ✅ Loaded embeddings for data_25_exponential (Model: t5) ✅ Loaded embeddings for data_94_quadratic (Model: bert) ✅ Loaded embeddings for data_94_quadratic (Model: ernie) ✅ Loaded embeddings for data_94_quadratic (Model: gatortron) ✅ Loaded embeddings for data_94_quadratic (Model: roberta) ✅ Loaded embeddings for data_94_quadratic (Model: t5) ✅ Loaded embeddings for data_96_cubic (Model: bert) ✅ Loaded embeddings for data_96_cubic (Model: ernie) ✅ Loaded embeddings for data_96_cubic (Model: gatortron) ✅ Loaded embeddings for data_96_cubic (Model: roberta) ✅ Loaded embeddings for data_96_cubic (Model: t5) ✅ Loaded embeddings for data_85_quadratic (Model: bert) ✅ Loaded embeddings for data_85_quadratic (Model: ernie) ✅ Loaded embeddings for data_85_quadratic (Model: gatortron) ✅ Loaded embeddings for data_85_quadratic (Model: roberta) ✅ Loaded embeddings for data_85_quadratic (Model: t5) ✅ Loaded embeddings for data_51_quadratic (Model: bert) ✅ Loaded embeddings for data_51_quadratic (Model: ernie) ✅ Loaded embeddings for data_51_quadratic (Model: gatortron) ✅ Loaded embeddings for data_51_quadratic (Model: roberta) ✅ Loaded embeddings for data_51_quadratic (Model: t5) ✅ Loaded embeddings for data_90_quadratic (Model: bert) ✅ Loaded embeddings for data_90_quadratic (Model: ernie) ✅ Loaded embeddings for data_90_quadratic (Model: gatortron) ✅ Loaded embeddings for data_90_quadratic (Model: roberta) ✅ Loaded embeddings for data_90_quadratic (Model: t5) ✅ Loaded embeddings for data_38_exponential (Model: bert) ✅ Loaded embeddings for data_38_exponential (Model: ernie) ✅ Loaded embeddings for data_38_exponential (Model: gatortron) ✅ Loaded embeddings for data_38_exponential (Model: roberta) ✅ Loaded embeddings for data_38_exponential (Model: t5) ✅ Loaded embeddings for data_81_exponential (Model: bert) ✅ Loaded embeddings for data_81_exponential (Model: ernie) ✅ Loaded embeddings for data_81_exponential (Model: gatortron) ✅ Loaded embeddings for data_81_exponential (Model: roberta) ✅ Loaded embeddings for data_81_exponential (Model: t5) ✅ Loaded embeddings for data_92_exponential (Model: bert) ✅ Loaded embeddings for data_92_exponential (Model: ernie) ✅ Loaded embeddings for data_92_exponential (Model: gatortron) ✅ Loaded embeddings for data_92_exponential (Model: roberta) ✅ Loaded embeddings for data_92_exponential (Model: t5) ✅ Loaded embeddings for data_62_quadratic (Model: bert) ✅ Loaded embeddings for data_62_quadratic (Model: ernie) ✅ Loaded embeddings for data_62_quadratic (Model: gatortron) ✅ Loaded embeddings for data_62_quadratic (Model: roberta) ✅ Loaded embeddings for data_62_quadratic (Model: t5) ✅ Loaded embeddings for data_10_linear (Model: bert) ✅ Loaded embeddings for data_10_linear (Model: ernie) ✅ Loaded embeddings for data_10_linear (Model: gatortron) ✅ Loaded embeddings for data_10_linear (Model: roberta) ✅ Loaded embeddings for data_10_linear (Model: t5) ✅ Loaded embeddings for data_31_linear (Model: bert) ✅ Loaded embeddings for data_31_linear (Model: ernie) ✅ Loaded embeddings for data_31_linear (Model: gatortron) ✅ Loaded embeddings for data_31_linear (Model: roberta) ✅ Loaded embeddings for data_31_linear (Model: t5) ✅ Loaded embeddings for data_57_linear (Model: bert) ✅ Loaded embeddings for data_57_linear (Model: ernie) ✅ Loaded embeddings for data_57_linear (Model: gatortron) ✅ Loaded embeddings for data_57_linear (Model: roberta) ✅ Loaded embeddings for data_57_linear (Model: t5) ✅ Loaded embeddings for data_70_linear (Model: bert) ✅ Loaded embeddings for data_70_linear (Model: ernie) ✅ Loaded embeddings for data_70_linear (Model: gatortron) ✅ Loaded embeddings for data_70_linear (Model: roberta) ✅ Loaded embeddings for data_70_linear (Model: t5) ✅ Loaded embeddings for data_98_cubic (Model: bert) ✅ Loaded embeddings for data_98_cubic (Model: ernie) ✅ Loaded embeddings for data_98_cubic (Model: gatortron) ✅ Loaded embeddings for data_98_cubic (Model: roberta) ✅ Loaded embeddings for data_98_cubic (Model: t5) ✅ Loaded embeddings for data_23_cubic (Model: bert) ✅ Loaded embeddings for data_23_cubic (Model: ernie) ✅ Loaded embeddings for data_23_cubic (Model: gatortron) ✅ Loaded embeddings for data_23_cubic (Model: roberta) ✅ Loaded embeddings for data_23_cubic (Model: t5) ✅ Loaded embeddings for data_27_exponential (Model: bert) ✅ Loaded embeddings for data_27_exponential (Model: ernie) ✅ Loaded embeddings for data_27_exponential (Model: gatortron) ✅ Loaded embeddings for data_27_exponential (Model: roberta) ✅ Loaded embeddings for data_27_exponential (Model: t5) ✅ Loaded embeddings for data_61_exponential (Model: bert) ✅ Loaded embeddings for data_61_exponential (Model: ernie) ✅ Loaded embeddings for data_61_exponential (Model: gatortron) ✅ Loaded embeddings for data_61_exponential (Model: roberta) ✅ Loaded embeddings for data_61_exponential (Model: t5) ✅ Loaded embeddings for data_46_linear (Model: bert) ✅ Loaded embeddings for data_46_linear (Model: ernie) ✅ Loaded embeddings for data_46_linear (Model: gatortron) ✅ Loaded embeddings for data_46_linear (Model: roberta) ✅ Loaded embeddings for data_46_linear (Model: t5) ✅ Loaded embeddings for data_73_cubic (Model: bert) ✅ Loaded embeddings for data_73_cubic (Model: ernie) ✅ Loaded embeddings for data_73_cubic (Model: gatortron) ✅ Loaded embeddings for data_73_cubic (Model: roberta) ✅ Loaded embeddings for data_73_cubic (Model: t5) ✅ Loaded embeddings for data_53_quadratic (Model: bert) ✅ Loaded embeddings for data_53_quadratic (Model: ernie) ✅ Loaded embeddings for data_53_quadratic (Model: gatortron) ✅ Loaded embeddings for data_53_quadratic (Model: roberta) ✅ Loaded embeddings for data_53_quadratic (Model: t5) ✅ Loaded embeddings for data_37_linear (Model: bert) ✅ Loaded embeddings for data_37_linear (Model: ernie) ✅ Loaded embeddings for data_37_linear (Model: gatortron) ✅ Loaded embeddings for data_37_linear (Model: roberta) ✅ Loaded embeddings for data_37_linear (Model: t5) ✅ Loaded embeddings for data_54_quadratic (Model: bert) ✅ Loaded embeddings for data_54_quadratic (Model: ernie) ✅ Loaded embeddings for data_54_quadratic (Model: gatortron) ✅ Loaded embeddings for data_54_quadratic (Model: roberta) ✅ Loaded embeddings for data_54_quadratic (Model: t5) ✅ Loaded embeddings for data_20_quadratic (Model: bert) ✅ Loaded embeddings for data_20_quadratic (Model: ernie) ✅ Loaded embeddings for data_20_quadratic (Model: gatortron) ✅ Loaded embeddings for data_20_quadratic (Model: roberta) ✅ Loaded embeddings for data_20_quadratic (Model: t5) ✅ Loaded embeddings for data_78_quadratic (Model: bert) ✅ Loaded embeddings for data_78_quadratic (Model: ernie) ✅ Loaded embeddings for data_78_quadratic (Model: gatortron) ✅ Loaded embeddings for data_78_quadratic (Model: roberta) ✅ Loaded embeddings for data_78_quadratic (Model: t5) ✅ Loaded embeddings for data_82_quadratic (Model: bert) ✅ Loaded embeddings for data_82_quadratic (Model: ernie) ✅ Loaded embeddings for data_82_quadratic (Model: gatortron) ✅ Loaded embeddings for data_82_quadratic (Model: roberta) ✅ Loaded embeddings for data_82_quadratic (Model: t5) ✅ Loaded embeddings for data_19_linear (Model: bert) ✅ Loaded embeddings for data_19_linear (Model: ernie) ✅ Loaded embeddings for data_19_linear (Model: gatortron) ✅ Loaded embeddings for data_19_linear (Model: roberta) ✅ Loaded embeddings for data_19_linear (Model: t5) ✅ Loaded embeddings for data_4_quadratic (Model: bert) ✅ Loaded embeddings for data_4_quadratic (Model: ernie) ✅ Loaded embeddings for data_4_quadratic (Model: gatortron) ✅ Loaded embeddings for data_4_quadratic (Model: roberta) ✅ Loaded embeddings for data_4_quadratic (Model: t5) ✅ Loaded embeddings for data_8_quadratic (Model: bert) ✅ Loaded embeddings for data_8_quadratic (Model: ernie) ✅ Loaded embeddings for data_8_quadratic (Model: gatortron) ✅ Loaded embeddings for data_8_quadratic (Model: roberta) ✅ Loaded embeddings for data_8_quadratic (Model: t5) ✅ Loaded embeddings for data_36_cubic (Model: bert) ✅ Loaded embeddings for data_36_cubic (Model: ernie) ✅ Loaded embeddings for data_36_cubic (Model: gatortron) ✅ Loaded embeddings for data_36_cubic (Model: roberta) ✅ Loaded embeddings for data_36_cubic (Model: t5) ✅ Loaded embeddings for data_24_exponential (Model: bert) ✅ Loaded embeddings for data_24_exponential (Model: ernie) ✅ Loaded embeddings for data_24_exponential (Model: gatortron) ✅ Loaded embeddings for data_24_exponential (Model: roberta) ✅ Loaded embeddings for data_24_exponential (Model: t5) ✅ Loaded embeddings for data_39_quadratic (Model: bert) ✅ Loaded embeddings for data_39_quadratic (Model: ernie) ✅ Loaded embeddings for data_39_quadratic (Model: gatortron) ✅ Loaded embeddings for data_39_quadratic (Model: roberta) ✅ Loaded embeddings for data_39_quadratic (Model: t5) ✅ Loaded embeddings for data_60_cubic (Model: bert) ✅ Loaded embeddings for data_60_cubic (Model: ernie) ✅ Loaded embeddings for data_60_cubic (Model: gatortron) ✅ Loaded embeddings for data_60_cubic (Model: roberta) ✅ Loaded embeddings for data_60_cubic (Model: t5) ✅ Loaded embeddings for data_1_quadratic (Model: bert) ✅ Loaded embeddings for data_1_quadratic (Model: ernie) ✅ Loaded embeddings for data_1_quadratic (Model: gatortron) ✅ Loaded embeddings for data_1_quadratic (Model: roberta) ✅ Loaded embeddings for data_1_quadratic (Model: t5) ✅ Loaded embeddings for data_41_exponential (Model: bert) ✅ Loaded embeddings for data_41_exponential (Model: ernie) ✅ Loaded embeddings for data_41_exponential (Model: gatortron) ✅ Loaded embeddings for data_41_exponential (Model: roberta) ✅ Loaded embeddings for data_41_exponential (Model: t5) ✅ Loaded embeddings for data_32_linear (Model: bert) ✅ Loaded embeddings for data_32_linear (Model: ernie) ✅ Loaded embeddings for data_32_linear (Model: gatortron) ✅ Loaded embeddings for data_32_linear (Model: roberta) ✅ Loaded embeddings for data_32_linear (Model: t5) ✅ Loaded embeddings for data_66_linear (Model: bert) ✅ Loaded embeddings for data_66_linear (Model: ernie) ✅ Loaded embeddings for data_66_linear (Model: gatortron) ✅ Loaded embeddings for data_66_linear (Model: roberta) ✅ Loaded embeddings for data_66_linear (Model: t5) ✅ Loaded embeddings for data_49_linear (Model: bert) ✅ Loaded embeddings for data_49_linear (Model: ernie) ✅ Loaded embeddings for data_49_linear (Model: gatortron) ✅ Loaded embeddings for data_49_linear (Model: roberta) ✅ Loaded embeddings for data_49_linear (Model: t5) ✅ Loaded embeddings for data_56_linear (Model: bert) ✅ Loaded embeddings for data_56_linear (Model: ernie) ✅ Loaded embeddings for data_56_linear (Model: gatortron) ✅ Loaded embeddings for data_56_linear (Model: roberta) ✅ Loaded embeddings for data_56_linear (Model: t5) ✅ Loaded embeddings for data_63_exponential (Model: bert) ✅ Loaded embeddings for data_63_exponential (Model: ernie) ✅ Loaded embeddings for data_63_exponential (Model: gatortron) ✅ Loaded embeddings for data_63_exponential (Model: roberta) ✅ Loaded embeddings for data_63_exponential (Model: t5) ✅ Loaded embeddings for data_80_cubic (Model: bert) ✅ Loaded embeddings for data_80_cubic (Model: ernie) ✅ Loaded embeddings for data_80_cubic (Model: gatortron) ✅ Loaded embeddings for data_80_cubic (Model: roberta) ✅ Loaded embeddings for data_80_cubic (Model: t5) ✅ Loaded embeddings for data_12_exponential (Model: bert) ✅ Loaded embeddings for data_12_exponential (Model: ernie) ✅ Loaded embeddings for data_12_exponential (Model: gatortron) ✅ Loaded embeddings for data_12_exponential (Model: roberta) ✅ Loaded embeddings for data_12_exponential (Model: t5) ✅ Loaded embeddings for data_3_quadratic (Model: bert) ✅ Loaded embeddings for data_3_quadratic (Model: ernie) ✅ Loaded embeddings for data_3_quadratic (Model: gatortron) ✅ Loaded embeddings for data_3_quadratic (Model: roberta) ✅ Loaded embeddings for data_3_quadratic (Model: t5) ✅ Loaded embeddings for data_64_exponential (Model: bert) ✅ Loaded embeddings for data_64_exponential (Model: ernie) ✅ Loaded embeddings for data_64_exponential (Model: gatortron) ✅ Loaded embeddings for data_64_exponential (Model: roberta) ✅ Loaded embeddings for data_64_exponential (Model: t5) ✅ Loaded embeddings for data_35_cubic (Model: bert) ✅ Loaded embeddings for data_35_cubic (Model: ernie) ✅ Loaded embeddings for data_35_cubic (Model: gatortron) ✅ Loaded embeddings for data_35_cubic (Model: roberta) ✅ Loaded embeddings for data_35_cubic (Model: t5) ✅ Loaded embeddings for data_6_linear (Model: bert) ✅ Loaded embeddings for data_6_linear (Model: ernie) ✅ Loaded embeddings for data_6_linear (Model: gatortron) ✅ Loaded embeddings for data_6_linear (Model: roberta) ✅ Loaded embeddings for data_6_linear (Model: t5) ✅ Loaded embeddings for data_93_exponential (Model: bert) ✅ Loaded embeddings for data_93_exponential (Model: ernie) ✅ Loaded embeddings for data_93_exponential (Model: gatortron) ✅ Loaded embeddings for data_93_exponential (Model: roberta) ✅ Loaded embeddings for data_93_exponential (Model: t5) ✅ Loaded embeddings for data_29_linear (Model: bert) ✅ Loaded embeddings for data_29_linear (Model: ernie) ✅ Loaded embeddings for data_29_linear (Model: gatortron) ✅ Loaded embeddings for data_29_linear (Model: roberta) ✅ Loaded embeddings for data_29_linear (Model: t5) ✅ Loaded embeddings for data_76_cubic (Model: bert) ✅ Loaded embeddings for data_76_cubic (Model: ernie) ✅ Loaded embeddings for data_76_cubic (Model: gatortron) ✅ Loaded embeddings for data_76_cubic (Model: roberta) ✅ Loaded embeddings for data_76_cubic (Model: t5) ✅ Loaded embeddings for data_50_exponential (Model: bert) ✅ Loaded embeddings for data_50_exponential (Model: ernie) ✅ Loaded embeddings for data_50_exponential (Model: gatortron) ✅ Loaded embeddings for data_50_exponential (Model: roberta) ✅ Loaded embeddings for data_50_exponential (Model: t5) ✅ Loaded embeddings for data_9_exponential (Model: bert) ✅ Loaded embeddings for data_9_exponential (Model: ernie) ✅ Loaded embeddings for data_9_exponential (Model: gatortron) ✅ Loaded embeddings for data_9_exponential (Model: roberta) ✅ Loaded embeddings for data_9_exponential (Model: t5) ✅ Loaded embeddings for data_59_quadratic (Model: bert) ✅ Loaded embeddings for data_59_quadratic (Model: ernie) ✅ Loaded embeddings for data_59_quadratic (Model: gatortron) ✅ Loaded embeddings for data_59_quadratic (Model: roberta) ✅ Loaded embeddings for data_59_quadratic (Model: t5) ✅ Loaded embeddings for data_97_quadratic (Model: bert) ✅ Loaded embeddings for data_97_quadratic (Model: ernie) ✅ Loaded embeddings for data_97_quadratic (Model: gatortron) ✅ Loaded embeddings for data_97_quadratic (Model: roberta) ✅ Loaded embeddings for data_97_quadratic (Model: t5) ✅ Loaded embeddings for data_45_cubic (Model: bert) ✅ Loaded embeddings for data_45_cubic (Model: ernie) ✅ Loaded embeddings for data_45_cubic (Model: gatortron) ✅ Loaded embeddings for data_45_cubic (Model: roberta) ✅ Loaded embeddings for data_45_cubic (Model: t5) ✅ Loaded embeddings for data_18_quadratic (Model: bert) ✅ Loaded embeddings for data_18_quadratic (Model: ernie) ✅ Loaded embeddings for data_18_quadratic (Model: gatortron) ✅ Loaded embeddings for data_18_quadratic (Model: roberta) ✅ Loaded embeddings for data_18_quadratic (Model: t5) ✅ Loaded embeddings for data_69_linear (Model: bert) ✅ Loaded embeddings for data_69_linear (Model: ernie) ✅ Loaded embeddings for data_69_linear (Model: gatortron) ✅ Loaded embeddings for data_69_linear (Model: roberta) ✅ Loaded embeddings for data_69_linear (Model: t5) ✅ Loaded embeddings for data_67_linear (Model: bert) ✅ Loaded embeddings for data_67_linear (Model: ernie) ✅ Loaded embeddings for data_67_linear (Model: gatortron) ✅ Loaded embeddings for data_67_linear (Model: roberta) ✅ Loaded embeddings for data_67_linear (Model: t5) ✅ Loaded embeddings for data_58_quadratic (Model: bert) ✅ Loaded embeddings for data_58_quadratic (Model: ernie) ✅ Loaded embeddings for data_58_quadratic (Model: gatortron) ✅ Loaded embeddings for data_58_quadratic (Model: roberta) ✅ Loaded embeddings for data_58_quadratic (Model: t5) ✅ Loaded embeddings for data_95_quadratic (Model: bert) ✅ Loaded embeddings for data_95_quadratic (Model: ernie) ✅ Loaded embeddings for data_95_quadratic (Model: gatortron) ✅ Loaded embeddings for data_95_quadratic (Model: roberta) ✅ Loaded embeddings for data_95_quadratic (Model: t5) ✅ Loaded embeddings for data_11_exponential (Model: bert) ✅ Loaded embeddings for data_11_exponential (Model: ernie) ✅ Loaded embeddings for data_11_exponential (Model: gatortron) ✅ Loaded embeddings for data_11_exponential (Model: roberta) ✅ Loaded embeddings for data_11_exponential (Model: t5) ✅ Loaded embeddings for data_87_quadratic (Model: bert) ✅ Loaded embeddings for data_87_quadratic (Model: ernie) ✅ Loaded embeddings for data_87_quadratic (Model: gatortron) ✅ Loaded embeddings for data_87_quadratic (Model: roberta) ✅ Loaded embeddings for data_87_quadratic (Model: t5) ✅ Loaded embeddings for data_2_linear (Model: bert) ✅ Loaded embeddings for data_2_linear (Model: ernie) ✅ Loaded embeddings for data_2_linear (Model: gatortron) ✅ Loaded embeddings for data_2_linear (Model: roberta) ✅ Loaded embeddings for data_2_linear (Model: t5) ✅ Loaded embeddings for data_47_cubic (Model: bert) ✅ Loaded embeddings for data_47_cubic (Model: ernie) ✅ Loaded embeddings for data_47_cubic (Model: gatortron) ✅ Loaded embeddings for data_47_cubic (Model: roberta) ✅ Loaded embeddings for data_47_cubic (Model: t5) ✅ Loaded embeddings for data_75_exponential (Model: bert) ✅ Loaded embeddings for data_75_exponential (Model: ernie) ✅ Loaded embeddings for data_75_exponential (Model: gatortron) ✅ Loaded embeddings for data_75_exponential (Model: roberta) ✅ Loaded embeddings for data_75_exponential (Model: t5) ✅ Loaded embeddings for data_83_exponential (Model: bert) ✅ Loaded embeddings for data_83_exponential (Model: ernie) ✅ Loaded embeddings for data_83_exponential (Model: gatortron) ✅ Loaded embeddings for data_83_exponential (Model: roberta) ✅ Loaded embeddings for data_83_exponential (Model: t5) ✅ Loaded embeddings for data_99_exponential (Model: bert) ✅ Loaded embeddings for data_99_exponential (Model: ernie) ✅ Loaded embeddings for data_99_exponential (Model: gatortron) ✅ Loaded embeddings for data_99_exponential (Model: roberta) ✅ Loaded embeddings for data_99_exponential (Model: t5) ✅ Loaded embeddings for data_7_cubic (Model: bert) ✅ Loaded embeddings for data_7_cubic (Model: ernie) ✅ Loaded embeddings for data_7_cubic (Model: gatortron) ✅ Loaded embeddings for data_7_cubic (Model: roberta) ✅ Loaded embeddings for data_7_cubic (Model: t5) ✅ Loaded embeddings for data_15_quadratic (Model: bert) ✅ Loaded embeddings for data_15_quadratic (Model: ernie) ✅ Loaded embeddings for data_15_quadratic (Model: gatortron) ✅ Loaded embeddings for data_15_quadratic (Model: roberta) ✅ Loaded embeddings for data_15_quadratic (Model: t5) ✅ Loaded embeddings for data_71_cubic (Model: bert) ✅ Loaded embeddings for data_71_cubic (Model: ernie) ✅ Loaded embeddings for data_71_cubic (Model: gatortron) ✅ Loaded embeddings for data_71_cubic (Model: roberta) ✅ Loaded embeddings for data_71_cubic (Model: t5) ✅ Loaded embeddings for data_77_cubic (Model: bert) ✅ Loaded embeddings for data_77_cubic (Model: ernie) ✅ Loaded embeddings for data_77_cubic (Model: gatortron) ✅ Loaded embeddings for data_77_cubic (Model: roberta) ✅ Loaded embeddings for data_77_cubic (Model: t5) ✅ Loaded embeddings for data_86_linear (Model: bert) ✅ Loaded embeddings for data_86_linear (Model: ernie) ✅ Loaded embeddings for data_86_linear (Model: gatortron) ✅ Loaded embeddings for data_86_linear (Model: roberta) ✅ Loaded embeddings for data_86_linear (Model: t5) ✅ Loaded embeddings for data_26_quadratic (Model: bert) ✅ Loaded embeddings for data_26_quadratic (Model: ernie) ✅ Loaded embeddings for data_26_quadratic (Model: gatortron) ✅ Loaded embeddings for data_26_quadratic (Model: roberta) ✅ Loaded embeddings for data_26_quadratic (Model: t5) ✅ Loaded embeddings for data_28_linear (Model: bert) ✅ Loaded embeddings for data_28_linear (Model: ernie) ✅ Loaded embeddings for data_28_linear (Model: gatortron) ✅ Loaded embeddings for data_28_linear (Model: roberta) ✅ Loaded embeddings for data_28_linear (Model: t5) 📊 Processing: data_79_quadratic (Model: bert) 📊 Processing: data_79_quadratic (Model: ernie) 📊 Processing: data_79_quadratic (Model: gatortron) 📊 Processing: data_79_quadratic (Model: roberta) 📊 Processing: data_79_quadratic (Model: t5) 📊 Processing: data_91_linear (Model: bert) 📊 Processing: data_91_linear (Model: ernie) 📊 Processing: data_91_linear (Model: gatortron) 📊 Processing: data_91_linear (Model: roberta) 📊 Processing: data_91_linear (Model: t5) 📊 Processing: data_44_exponential (Model: bert) 📊 Processing: data_44_exponential (Model: ernie) 📊 Processing: data_44_exponential (Model: gatortron) 📊 Processing: data_44_exponential (Model: roberta) 📊 Processing: data_44_exponential (Model: t5) 📊 Processing: data_42_linear (Model: bert) 📊 Processing: data_42_linear (Model: ernie) 📊 Processing: data_42_linear (Model: gatortron) 📊 Processing: data_42_linear (Model: roberta) 📊 Processing: data_42_linear (Model: t5) 📊 Processing: data_88_exponential (Model: bert) 📊 Processing: data_88_exponential (Model: ernie) 📊 Processing: data_88_exponential (Model: gatortron) 📊 Processing: data_88_exponential (Model: roberta) 📊 Processing: data_88_exponential (Model: t5) 📊 Processing: data_43_linear (Model: bert) 📊 Processing: data_43_linear (Model: ernie) 📊 Processing: data_43_linear (Model: gatortron) 📊 Processing: data_43_linear (Model: roberta) 📊 Processing: data_43_linear (Model: t5) 📊 Processing: data_68_exponential (Model: bert) 📊 Processing: data_68_exponential (Model: ernie) 📊 Processing: data_68_exponential (Model: gatortron) 📊 Processing: data_68_exponential (Model: roberta) 📊 Processing: data_68_exponential (Model: t5) 📊 Processing: data_48_cubic (Model: bert) 📊 Processing: data_48_cubic (Model: ernie) 📊 Processing: data_48_cubic (Model: gatortron) 📊 Processing: data_48_cubic (Model: roberta) 📊 Processing: data_48_cubic (Model: t5) 📊 Processing: data_13_cubic (Model: bert) 📊 Processing: data_13_cubic (Model: ernie) 📊 Processing: data_13_cubic (Model: gatortron) 📊 Processing: data_13_cubic (Model: roberta) 📊 Processing: data_13_cubic (Model: t5) 📊 Processing: data_33_quadratic (Model: bert) 📊 Processing: data_33_quadratic (Model: ernie) 📊 Processing: data_33_quadratic (Model: gatortron) 📊 Processing: data_33_quadratic (Model: roberta) 📊 Processing: data_33_quadratic (Model: t5) 📊 Processing: data_84_cubic (Model: bert) 📊 Processing: data_84_cubic (Model: ernie) 📊 Processing: data_84_cubic (Model: gatortron) 📊 Processing: data_84_cubic (Model: roberta) 📊 Processing: data_84_cubic (Model: t5) 📊 Processing: data_74_linear (Model: bert) 📊 Processing: data_74_linear (Model: ernie) 📊 Processing: data_74_linear (Model: gatortron) 📊 Processing: data_74_linear (Model: roberta) 📊 Processing: data_74_linear (Model: t5) 📊 Processing: data_21_quadratic (Model: bert) 📊 Processing: data_21_quadratic (Model: ernie) 📊 Processing: data_21_quadratic (Model: gatortron) 📊 Processing: data_21_quadratic (Model: roberta) 📊 Processing: data_21_quadratic (Model: t5) 📊 Processing: data_40_exponential (Model: bert) 📊 Processing: data_40_exponential (Model: ernie) 📊 Processing: data_40_exponential (Model: gatortron) 📊 Processing: data_40_exponential (Model: roberta) 📊 Processing: data_40_exponential (Model: t5) 📊 Processing: data_5_cubic (Model: bert) 📊 Processing: data_5_cubic (Model: ernie) 📊 Processing: data_5_cubic (Model: gatortron) 📊 Processing: data_5_cubic (Model: roberta) 📊 Processing: data_5_cubic (Model: t5) 📊 Processing: data_17_exponential (Model: bert) 📊 Processing: data_17_exponential (Model: ernie) 📊 Processing: data_17_exponential (Model: gatortron) 📊 Processing: data_17_exponential (Model: roberta) 📊 Processing: data_17_exponential (Model: t5) 📊 Processing: data_0_exponential (Model: bert) 📊 Processing: data_0_exponential (Model: ernie) 📊 Processing: data_0_exponential (Model: gatortron) 📊 Processing: data_0_exponential (Model: roberta) 📊 Processing: data_0_exponential (Model: t5) 📊 Processing: data_30_cubic (Model: bert) 📊 Processing: data_30_cubic (Model: ernie) 📊 Processing: data_30_cubic (Model: gatortron) 📊 Processing: data_30_cubic (Model: roberta) 📊 Processing: data_30_cubic (Model: t5) 📊 Processing: data_34_cubic (Model: bert) 📊 Processing: data_34_cubic (Model: ernie) 📊 Processing: data_34_cubic (Model: gatortron) 📊 Processing: data_34_cubic (Model: roberta) 📊 Processing: data_34_cubic (Model: t5) 📊 Processing: data_52_cubic (Model: bert) 📊 Processing: data_52_cubic (Model: ernie) 📊 Processing: data_52_cubic (Model: gatortron) 📊 Processing: data_52_cubic (Model: roberta) 📊 Processing: data_52_cubic (Model: t5) 📊 Processing: data_16_cubic (Model: bert) 📊 Processing: data_16_cubic (Model: ernie) 📊 Processing: data_16_cubic (Model: gatortron) 📊 Processing: data_16_cubic (Model: roberta) 📊 Processing: data_16_cubic (Model: t5) 📊 Processing: data_72_linear (Model: bert) 📊 Processing: data_72_linear (Model: ernie) 📊 Processing: data_72_linear (Model: gatortron) 📊 Processing: data_72_linear (Model: roberta) 📊 Processing: data_72_linear (Model: t5) 📊 Processing: data_89_linear (Model: bert) 📊 Processing: data_89_linear (Model: ernie) 📊 Processing: data_89_linear (Model: gatortron) 📊 Processing: data_89_linear (Model: roberta) 📊 Processing: data_89_linear (Model: t5) 📊 Processing: data_55_linear (Model: bert) 📊 Processing: data_55_linear (Model: ernie) 📊 Processing: data_55_linear (Model: gatortron) 📊 Processing: data_55_linear (Model: roberta) 📊 Processing: data_55_linear (Model: t5) 📊 Processing: data_22_quadratic (Model: bert) 📊 Processing: data_22_quadratic (Model: ernie) 📊 Processing: data_22_quadratic (Model: gatortron) 📊 Processing: data_22_quadratic (Model: roberta) 📊 Processing: data_22_quadratic (Model: t5) 📊 Processing: data_65_exponential (Model: bert) 📊 Processing: data_65_exponential (Model: ernie) 📊 Processing: data_65_exponential (Model: gatortron) 📊 Processing: data_65_exponential (Model: roberta) 📊 Processing: data_65_exponential (Model: t5) 📊 Processing: data_14_linear (Model: bert) 📊 Processing: data_14_linear (Model: ernie) 📊 Processing: data_14_linear (Model: gatortron) 📊 Processing: data_14_linear (Model: roberta) 📊 Processing: data_14_linear (Model: t5) 📊 Processing: data_25_exponential (Model: bert) 📊 Processing: data_25_exponential (Model: ernie) 📊 Processing: data_25_exponential (Model: gatortron) 📊 Processing: data_25_exponential (Model: roberta) 📊 Processing: data_25_exponential (Model: t5) 📊 Processing: data_94_quadratic (Model: bert) 📊 Processing: data_94_quadratic (Model: ernie) 📊 Processing: data_94_quadratic (Model: gatortron) 📊 Processing: data_94_quadratic (Model: roberta) 📊 Processing: data_94_quadratic (Model: t5) 📊 Processing: data_96_cubic (Model: bert) 📊 Processing: data_96_cubic (Model: ernie) 📊 Processing: data_96_cubic (Model: gatortron) 📊 Processing: data_96_cubic (Model: roberta) 📊 Processing: data_96_cubic (Model: t5) 📊 Processing: data_85_quadratic (Model: bert) 📊 Processing: data_85_quadratic (Model: ernie) 📊 Processing: data_85_quadratic (Model: gatortron) 📊 Processing: data_85_quadratic (Model: roberta) 📊 Processing: data_85_quadratic (Model: t5) 📊 Processing: data_51_quadratic (Model: bert) 📊 Processing: data_51_quadratic (Model: ernie) 📊 Processing: data_51_quadratic (Model: gatortron) 📊 Processing: data_51_quadratic (Model: roberta) 📊 Processing: data_51_quadratic (Model: t5) 📊 Processing: data_90_quadratic (Model: bert) 📊 Processing: data_90_quadratic (Model: ernie) 📊 Processing: data_90_quadratic (Model: gatortron) 📊 Processing: data_90_quadratic (Model: roberta) 📊 Processing: data_90_quadratic (Model: t5) 📊 Processing: data_38_exponential (Model: bert) 📊 Processing: data_38_exponential (Model: ernie) 📊 Processing: data_38_exponential (Model: gatortron) 📊 Processing: data_38_exponential (Model: roberta) 📊 Processing: data_38_exponential (Model: t5) 📊 Processing: data_81_exponential (Model: bert) 📊 Processing: data_81_exponential (Model: ernie) 📊 Processing: data_81_exponential (Model: gatortron) 📊 Processing: data_81_exponential (Model: roberta) 📊 Processing: data_81_exponential (Model: t5) 📊 Processing: data_92_exponential (Model: bert) 📊 Processing: data_92_exponential (Model: ernie) 📊 Processing: data_92_exponential (Model: gatortron) 📊 Processing: data_92_exponential (Model: roberta) 📊 Processing: data_92_exponential (Model: t5) 📊 Processing: data_62_quadratic (Model: bert) 📊 Processing: data_62_quadratic (Model: ernie) 📊 Processing: data_62_quadratic (Model: gatortron) 📊 Processing: data_62_quadratic (Model: roberta) 📊 Processing: data_62_quadratic (Model: t5) 📊 Processing: data_10_linear (Model: bert) 📊 Processing: data_10_linear (Model: ernie) 📊 Processing: data_10_linear (Model: gatortron) 📊 Processing: data_10_linear (Model: roberta) 📊 Processing: data_10_linear (Model: t5) 📊 Processing: data_31_linear (Model: bert) 📊 Processing: data_31_linear (Model: ernie) 📊 Processing: data_31_linear (Model: gatortron) 📊 Processing: data_31_linear (Model: roberta) 📊 Processing: data_31_linear (Model: t5) 📊 Processing: data_57_linear (Model: bert) 📊 Processing: data_57_linear (Model: ernie) 📊 Processing: data_57_linear (Model: gatortron) 📊 Processing: data_57_linear (Model: roberta) 📊 Processing: data_57_linear (Model: t5) 📊 Processing: data_70_linear (Model: bert) 📊 Processing: data_70_linear (Model: ernie) 📊 Processing: data_70_linear (Model: gatortron) 📊 Processing: data_70_linear (Model: roberta) 📊 Processing: data_70_linear (Model: t5) 📊 Processing: data_98_cubic (Model: bert) 📊 Processing: data_98_cubic (Model: ernie) 📊 Processing: data_98_cubic (Model: gatortron) 📊 Processing: data_98_cubic (Model: roberta) 📊 Processing: data_98_cubic (Model: t5) 📊 Processing: data_23_cubic (Model: bert) 📊 Processing: data_23_cubic (Model: ernie) 📊 Processing: data_23_cubic (Model: gatortron) 📊 Processing: data_23_cubic (Model: roberta) 📊 Processing: data_23_cubic (Model: t5) 📊 Processing: data_27_exponential (Model: bert) 📊 Processing: data_27_exponential (Model: ernie) 📊 Processing: data_27_exponential (Model: gatortron) 📊 Processing: data_27_exponential (Model: roberta) 📊 Processing: data_27_exponential (Model: t5) 📊 Processing: data_61_exponential (Model: bert) 📊 Processing: data_61_exponential (Model: ernie) 📊 Processing: data_61_exponential (Model: gatortron) 📊 Processing: data_61_exponential (Model: roberta) 📊 Processing: data_61_exponential (Model: t5) 📊 Processing: data_46_linear (Model: bert) 📊 Processing: data_46_linear (Model: ernie) 📊 Processing: data_46_linear (Model: gatortron) 📊 Processing: data_46_linear (Model: roberta) 📊 Processing: data_46_linear (Model: t5) 📊 Processing: data_73_cubic (Model: bert) 📊 Processing: data_73_cubic (Model: ernie) 📊 Processing: data_73_cubic (Model: gatortron) 📊 Processing: data_73_cubic (Model: roberta) 📊 Processing: data_73_cubic (Model: t5) 📊 Processing: data_53_quadratic (Model: bert) 📊 Processing: data_53_quadratic (Model: ernie) 📊 Processing: data_53_quadratic (Model: gatortron) 📊 Processing: data_53_quadratic (Model: roberta) 📊 Processing: data_53_quadratic (Model: t5) 📊 Processing: data_37_linear (Model: bert) 📊 Processing: data_37_linear (Model: ernie) 📊 Processing: data_37_linear (Model: gatortron) 📊 Processing: data_37_linear (Model: roberta) 📊 Processing: data_37_linear (Model: t5) 📊 Processing: data_54_quadratic (Model: bert) 📊 Processing: data_54_quadratic (Model: ernie) 📊 Processing: data_54_quadratic (Model: gatortron) 📊 Processing: data_54_quadratic (Model: roberta) 📊 Processing: data_54_quadratic (Model: t5) 📊 Processing: data_20_quadratic (Model: bert) 📊 Processing: data_20_quadratic (Model: ernie) 📊 Processing: data_20_quadratic (Model: gatortron) 📊 Processing: data_20_quadratic (Model: roberta) 📊 Processing: data_20_quadratic (Model: t5) 📊 Processing: data_78_quadratic (Model: bert) 📊 Processing: data_78_quadratic (Model: ernie) 📊 Processing: data_78_quadratic (Model: gatortron) 📊 Processing: data_78_quadratic (Model: roberta) 📊 Processing: data_78_quadratic (Model: t5) 📊 Processing: data_82_quadratic (Model: bert) 📊 Processing: data_82_quadratic (Model: ernie) 📊 Processing: data_82_quadratic (Model: gatortron) 📊 Processing: data_82_quadratic (Model: roberta) 📊 Processing: data_82_quadratic (Model: t5) 📊 Processing: data_19_linear (Model: bert) 📊 Processing: data_19_linear (Model: ernie) 📊 Processing: data_19_linear (Model: gatortron) 📊 Processing: data_19_linear (Model: roberta) 📊 Processing: data_19_linear (Model: t5) 📊 Processing: data_4_quadratic (Model: bert) 📊 Processing: data_4_quadratic (Model: ernie) 📊 Processing: data_4_quadratic (Model: gatortron) 📊 Processing: data_4_quadratic (Model: roberta) 📊 Processing: data_4_quadratic (Model: t5) 📊 Processing: data_8_quadratic (Model: bert) 📊 Processing: data_8_quadratic (Model: ernie) 📊 Processing: data_8_quadratic (Model: gatortron) 📊 Processing: data_8_quadratic (Model: roberta) 📊 Processing: data_8_quadratic (Model: t5) 📊 Processing: data_36_cubic (Model: bert) 📊 Processing: data_36_cubic (Model: ernie) 📊 Processing: data_36_cubic (Model: gatortron) 📊 Processing: data_36_cubic (Model: roberta) 📊 Processing: data_36_cubic (Model: t5) 📊 Processing: data_24_exponential (Model: bert) 📊 Processing: data_24_exponential (Model: ernie) 📊 Processing: data_24_exponential (Model: gatortron) 📊 Processing: data_24_exponential (Model: roberta) 📊 Processing: data_24_exponential (Model: t5) 📊 Processing: data_39_quadratic (Model: bert) 📊 Processing: data_39_quadratic (Model: ernie) 📊 Processing: data_39_quadratic (Model: gatortron) 📊 Processing: data_39_quadratic (Model: roberta) 📊 Processing: data_39_quadratic (Model: t5) 📊 Processing: data_60_cubic (Model: bert) 📊 Processing: data_60_cubic (Model: ernie) 📊 Processing: data_60_cubic (Model: gatortron) 📊 Processing: data_60_cubic (Model: roberta) 📊 Processing: data_60_cubic (Model: t5) 📊 Processing: data_1_quadratic (Model: bert) 📊 Processing: data_1_quadratic (Model: ernie) 📊 Processing: data_1_quadratic (Model: gatortron) 📊 Processing: data_1_quadratic (Model: roberta) 📊 Processing: data_1_quadratic (Model: t5) 📊 Processing: data_41_exponential (Model: bert) 📊 Processing: data_41_exponential (Model: ernie) 📊 Processing: data_41_exponential (Model: gatortron) 📊 Processing: data_41_exponential (Model: roberta) 📊 Processing: data_41_exponential (Model: t5) 📊 Processing: data_32_linear (Model: bert) 📊 Processing: data_32_linear (Model: ernie) 📊 Processing: data_32_linear (Model: gatortron) 📊 Processing: data_32_linear (Model: roberta) 📊 Processing: data_32_linear (Model: t5) 📊 Processing: data_66_linear (Model: bert) 📊 Processing: data_66_linear (Model: ernie) 📊 Processing: data_66_linear (Model: gatortron) 📊 Processing: data_66_linear (Model: roberta) 📊 Processing: data_66_linear (Model: t5) 📊 Processing: data_49_linear (Model: bert) 📊 Processing: data_49_linear (Model: ernie) 📊 Processing: data_49_linear (Model: gatortron) 📊 Processing: data_49_linear (Model: roberta) 📊 Processing: data_49_linear (Model: t5) 📊 Processing: data_56_linear (Model: bert) 📊 Processing: data_56_linear (Model: ernie) 📊 Processing: data_56_linear (Model: gatortron) 📊 Processing: data_56_linear (Model: roberta) 📊 Processing: data_56_linear (Model: t5) 📊 Processing: data_63_exponential (Model: bert) 📊 Processing: data_63_exponential (Model: ernie) 📊 Processing: data_63_exponential (Model: gatortron) 📊 Processing: data_63_exponential (Model: roberta) 📊 Processing: data_63_exponential (Model: t5) 📊 Processing: data_80_cubic (Model: bert) 📊 Processing: data_80_cubic (Model: ernie) 📊 Processing: data_80_cubic (Model: gatortron) 📊 Processing: data_80_cubic (Model: roberta) 📊 Processing: data_80_cubic (Model: t5) 📊 Processing: data_12_exponential (Model: bert) 📊 Processing: data_12_exponential (Model: ernie) 📊 Processing: data_12_exponential (Model: gatortron) 📊 Processing: data_12_exponential (Model: roberta) 📊 Processing: data_12_exponential (Model: t5) 📊 Processing: data_3_quadratic (Model: bert) 📊 Processing: data_3_quadratic (Model: ernie) 📊 Processing: data_3_quadratic (Model: gatortron) 📊 Processing: data_3_quadratic (Model: roberta) 📊 Processing: data_3_quadratic (Model: t5) 📊 Processing: data_64_exponential (Model: bert) 📊 Processing: data_64_exponential (Model: ernie) 📊 Processing: data_64_exponential (Model: gatortron) 📊 Processing: data_64_exponential (Model: roberta) 📊 Processing: data_64_exponential (Model: t5) 📊 Processing: data_35_cubic (Model: bert) 📊 Processing: data_35_cubic (Model: ernie) 📊 Processing: data_35_cubic (Model: gatortron) 📊 Processing: data_35_cubic (Model: roberta) 📊 Processing: data_35_cubic (Model: t5) 📊 Processing: data_6_linear (Model: bert) 📊 Processing: data_6_linear (Model: ernie) 📊 Processing: data_6_linear (Model: gatortron) 📊 Processing: data_6_linear (Model: roberta) 📊 Processing: data_6_linear (Model: t5) 📊 Processing: data_93_exponential (Model: bert) 📊 Processing: data_93_exponential (Model: ernie) 📊 Processing: data_93_exponential (Model: gatortron) 📊 Processing: data_93_exponential (Model: roberta) 📊 Processing: data_93_exponential (Model: t5) 📊 Processing: data_29_linear (Model: bert) 📊 Processing: data_29_linear (Model: ernie) 📊 Processing: data_29_linear (Model: gatortron) 📊 Processing: data_29_linear (Model: roberta) 📊 Processing: data_29_linear (Model: t5) 📊 Processing: data_76_cubic (Model: bert) 📊 Processing: data_76_cubic (Model: ernie) 📊 Processing: data_76_cubic (Model: gatortron) 📊 Processing: data_76_cubic (Model: roberta) 📊 Processing: data_76_cubic (Model: t5) 📊 Processing: data_50_exponential (Model: bert) 📊 Processing: data_50_exponential (Model: ernie) 📊 Processing: data_50_exponential (Model: gatortron) 📊 Processing: data_50_exponential (Model: roberta) 📊 Processing: data_50_exponential (Model: t5) 📊 Processing: data_9_exponential (Model: bert) 📊 Processing: data_9_exponential (Model: ernie) 📊 Processing: data_9_exponential (Model: gatortron) 📊 Processing: data_9_exponential (Model: roberta) 📊 Processing: data_9_exponential (Model: t5) 📊 Processing: data_59_quadratic (Model: bert) 📊 Processing: data_59_quadratic (Model: ernie) 📊 Processing: data_59_quadratic (Model: gatortron) 📊 Processing: data_59_quadratic (Model: roberta) 📊 Processing: data_59_quadratic (Model: t5) 📊 Processing: data_97_quadratic (Model: bert) 📊 Processing: data_97_quadratic (Model: ernie) 📊 Processing: data_97_quadratic (Model: gatortron) 📊 Processing: data_97_quadratic (Model: roberta) 📊 Processing: data_97_quadratic (Model: t5) 📊 Processing: data_45_cubic (Model: bert) 📊 Processing: data_45_cubic (Model: ernie) 📊 Processing: data_45_cubic (Model: gatortron) 📊 Processing: data_45_cubic (Model: roberta) 📊 Processing: data_45_cubic (Model: t5) 📊 Processing: data_18_quadratic (Model: bert) 📊 Processing: data_18_quadratic (Model: ernie) 📊 Processing: data_18_quadratic (Model: gatortron) 📊 Processing: data_18_quadratic (Model: roberta) 📊 Processing: data_18_quadratic (Model: t5) 📊 Processing: data_69_linear (Model: bert) 📊 Processing: data_69_linear (Model: ernie) 📊 Processing: data_69_linear (Model: gatortron) 📊 Processing: data_69_linear (Model: roberta) 📊 Processing: data_69_linear (Model: t5) 📊 Processing: data_67_linear (Model: bert) 📊 Processing: data_67_linear (Model: ernie) 📊 Processing: data_67_linear (Model: gatortron) 📊 Processing: data_67_linear (Model: roberta) 📊 Processing: data_67_linear (Model: t5) 📊 Processing: data_58_quadratic (Model: bert) 📊 Processing: data_58_quadratic (Model: ernie) 📊 Processing: data_58_quadratic (Model: gatortron) 📊 Processing: data_58_quadratic (Model: roberta) 📊 Processing: data_58_quadratic (Model: t5) 📊 Processing: data_95_quadratic (Model: bert) 📊 Processing: data_95_quadratic (Model: ernie) 📊 Processing: data_95_quadratic (Model: gatortron) 📊 Processing: data_95_quadratic (Model: roberta) 📊 Processing: data_95_quadratic (Model: t5) 📊 Processing: data_11_exponential (Model: bert) 📊 Processing: data_11_exponential (Model: ernie) 📊 Processing: data_11_exponential (Model: gatortron) 📊 Processing: data_11_exponential (Model: roberta) 📊 Processing: data_11_exponential (Model: t5) 📊 Processing: data_87_quadratic (Model: bert) 📊 Processing: data_87_quadratic (Model: ernie) 📊 Processing: data_87_quadratic (Model: gatortron) 📊 Processing: data_87_quadratic (Model: roberta) 📊 Processing: data_87_quadratic (Model: t5) 📊 Processing: data_2_linear (Model: bert) 📊 Processing: data_2_linear (Model: ernie) 📊 Processing: data_2_linear (Model: gatortron) 📊 Processing: data_2_linear (Model: roberta) 📊 Processing: data_2_linear (Model: t5) 📊 Processing: data_47_cubic (Model: bert) 📊 Processing: data_47_cubic (Model: ernie) 📊 Processing: data_47_cubic (Model: gatortron) 📊 Processing: data_47_cubic (Model: roberta) 📊 Processing: data_47_cubic (Model: t5) 📊 Processing: data_75_exponential (Model: bert) 📊 Processing: data_75_exponential (Model: ernie) 📊 Processing: data_75_exponential (Model: gatortron) 📊 Processing: data_75_exponential (Model: roberta) 📊 Processing: data_75_exponential (Model: t5) 📊 Processing: data_83_exponential (Model: bert) 📊 Processing: data_83_exponential (Model: ernie) 📊 Processing: data_83_exponential (Model: gatortron) 📊 Processing: data_83_exponential (Model: roberta) 📊 Processing: data_83_exponential (Model: t5) 📊 Processing: data_99_exponential (Model: bert) 📊 Processing: data_99_exponential (Model: ernie) 📊 Processing: data_99_exponential (Model: gatortron) 📊 Processing: data_99_exponential (Model: roberta) 📊 Processing: data_99_exponential (Model: t5) 📊 Processing: data_7_cubic (Model: bert) 📊 Processing: data_7_cubic (Model: ernie) 📊 Processing: data_7_cubic (Model: gatortron) 📊 Processing: data_7_cubic (Model: roberta) 📊 Processing: data_7_cubic (Model: t5) 📊 Processing: data_15_quadratic (Model: bert) 📊 Processing: data_15_quadratic (Model: ernie) 📊 Processing: data_15_quadratic (Model: gatortron) 📊 Processing: data_15_quadratic (Model: roberta) 📊 Processing: data_15_quadratic (Model: t5) 📊 Processing: data_71_cubic (Model: bert) 📊 Processing: data_71_cubic (Model: ernie) 📊 Processing: data_71_cubic (Model: gatortron) 📊 Processing: data_71_cubic (Model: roberta) 📊 Processing: data_71_cubic (Model: t5) 📊 Processing: data_77_cubic (Model: bert) 📊 Processing: data_77_cubic (Model: ernie) 📊 Processing: data_77_cubic (Model: gatortron) 📊 Processing: data_77_cubic (Model: roberta) 📊 Processing: data_77_cubic (Model: t5) 📊 Processing: data_86_linear (Model: bert) 📊 Processing: data_86_linear (Model: ernie) 📊 Processing: data_86_linear (Model: gatortron) 📊 Processing: data_86_linear (Model: roberta) 📊 Processing: data_86_linear (Model: t5) 📊 Processing: data_26_quadratic (Model: bert) 📊 Processing: data_26_quadratic (Model: ernie) 📊 Processing: data_26_quadratic (Model: gatortron) 📊 Processing: data_26_quadratic (Model: roberta) 📊 Processing: data_26_quadratic (Model: t5) 📊 Processing: data_28_linear (Model: bert) 📊 Processing: data_28_linear (Model: ernie) 📊 Processing: data_28_linear (Model: gatortron) 📊 Processing: data_28_linear (Model: roberta) 📊 Processing: data_28_linear (Model: t5) 📂 Processing /home/abbasali/synthetic data/generated_data_llama -> /home/abbasali/synthetic data/embeddings_llama ✅ Loaded 39 CSV datasets from /home/abbasali/synthetic data/generated_data_llama. ⚠️ WARNING: Missing embeddings for data_79_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_91_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_44_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_42_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_84_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_74_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_17_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_52_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_72_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_22_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_65_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_25_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_85_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_51_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_70_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_98_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_27_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_53_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_54_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_78_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_4_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_36_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_24_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_49_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_56_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_80_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_12_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_93_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_50_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_9_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_59_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_69_linear (Model: llama) ⚠️ WARNING: Missing embeddings for data_58_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_87_quadratic (Model: llama) ⚠️ WARNING: Missing embeddings for data_47_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_75_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_83_exponential (Model: llama) ⚠️ WARNING: Missing embeddings for data_7_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_15_quadratic (Model: llama) ✅ Results saved to cluster_similarity_results1.csv
In [37]:
import os
import numpy as np
import pandas as pd
import h5py
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE
from sklearn.impute import SimpleImputer
from scipy.spatial.distance import pdist, squareform
from scipy.stats import spearmanr
from sklearn.metrics.pairwise import cosine_similarity
# ✅ PREDEFINED SETTINGS (Set to False if you don't want to re-run)
RE_RUN = True # Change to False to skip clustering analysis
RE_RUN_LLAMA = True # Change to False to skip Llama processing
# ✅ Define dataset-embedding pairs with correct file extensions
processing_sets = [
{
"input_dir": "/home/abbasali/synthetic data/generated_data",
"output_dir": "/home/abbasali/synthetic data/embeddings",
"file_ext": ".npy",
"models": ["bert", "ernie", "gatortron", "roberta", "t5", "e5_small", "minilm"]
},
{
"input_dir": "/home/abbasali/synthetic data/generated_data_llama",
"output_dir": "/home/abbasali/synthetic data/embeddings_llama",
"file_ext": ".h5",
"models": ["llama"]
}
]
# ✅ Load previous results if available
RESULTS_FILE = "cluster_similarity_results.csv"
if os.path.exists(RESULTS_FILE) and not RE_RUN:
existing_results = pd.read_csv(RESULTS_FILE)
processed_files = set(zip(existing_results["Dataset"], existing_results["Model"]))
else:
existing_results = pd.DataFrame()
processed_files = set()
def load_data(input_dir, output_dir, file_ext, models):
"""Loads CSV datasets and dynamically finds embeddings in the output directory."""
print(f"\n📂 Processing {input_dir} -> {output_dir}")
if not os.path.exists(input_dir):
print(f"❌ ERROR: Input directory not found: {input_dir}")
return {}, {}
if not os.path.exists(output_dir):
print(f"❌ ERROR: Embeddings directory not found: {output_dir}")
return {}, {}
# ✅ Explicitly list all files in the embeddings directory
existing_files = set(os.listdir(output_dir))
if not existing_files:
print(f"⚠️ No embeddings found in {output_dir}")
csv_files = [f for f in os.listdir(input_dir) if f.endswith(".csv")]
if not csv_files:
print(f"⚠️ No CSV datasets found in {input_dir} ❌")
dataframes = {os.path.splitext(f)[0]: pd.read_csv(os.path.join(input_dir, f)) for f in csv_files}
print(f"✅ Loaded {len(dataframes)} CSV datasets from {input_dir}.")
embeddings_data = {}
for filename in csv_files:
dataset_name = os.path.splitext(filename)[0]
for model_name in models:
# ✅ Fix: Match naming conventions for different models
if model_name == "llama":
matched_files = [f for f in existing_files if dataset_name in f and f.endswith(".h5")]
elif model_name == "e5_small":
matched_files = [f for f in existing_files if f.startswith(f"embeddings_e5_small_{dataset_name}") and f.endswith(".npy")]
elif model_name == "minilm":
matched_files = [f for f in existing_files if f.startswith(f"embeddings_minilm_{dataset_name}") and f.endswith(".npy")]
else:
matched_files = [f for f in existing_files if f.startswith(f"embeddings_{model_name}_{dataset_name}") and f.endswith(".npy")]
if not matched_files:
print(f"⚠️ WARNING: Missing embeddings for {dataset_name} (Model: {model_name})")
continue
embeddings_filepath = os.path.join(output_dir, matched_files[0]) # Use first match
if (dataset_name, model_name) in processed_files:
print(f"⚠️ Skipping already processed file: {dataset_name} (Model: {model_name})")
continue
try:
if model_name == "llama":
with h5py.File(embeddings_filepath, 'r') as h5f:
embeddings = h5f['embeddings'][:]
else:
embeddings = np.load(embeddings_filepath)
if embeddings.size == 0:
print(f"❌ ERROR: Empty embeddings file detected: {embeddings_filepath}")
continue
if dataset_name not in embeddings_data:
embeddings_data[dataset_name] = {}
embeddings_data[dataset_name][model_name] = embeddings
print(f"✅ Loaded embeddings for {dataset_name} (Model: {model_name})")
except Exception as e:
print(f"❌ ERROR: Could not load {embeddings_filepath} - {e}")
if not embeddings_data:
print(f"⚠️ No embeddings found in {output_dir}")
return dataframes, embeddings_data
def compute_cluster_similarity(dataframes, embeddings_data):
"""Computes cluster relationships using PCA, t-SNE, and KMeans clustering."""
results = []
for dataset_name, df in dataframes.items():
if df.empty:
print(f"⚠️ Skipping empty dataset: {dataset_name}")
continue
print(f"\n🔍 Computing cluster similarity for: {dataset_name}")
continuous_features = df.iloc[:, :3].values
sample_size = min(300, continuous_features.shape[0])
continuous_sample = SimpleImputer(strategy="mean").fit_transform(continuous_features[:sample_size])
optimal_clusters = min(40, len(np.unique(continuous_sample, axis=0)))
optimal_clusters = max(2, optimal_clusters)
kmeans_orig = KMeans(n_clusters=optimal_clusters, random_state=42, n_init=10).fit(continuous_sample)
dist_orig = squareform(pdist(kmeans_orig.cluster_centers_, metric='euclidean'))
for model_name, embeddings in embeddings_data.get(dataset_name, {}).items():
print(f"📊 Processing embeddings for {dataset_name} (Model: {model_name})")
# ✅ Fix: Ensure embeddings are 2D
if embeddings.ndim > 2:
print(f"⚠️ Warning: Found 3D embeddings for {dataset_name} (Model: {model_name}), flattening...")
embeddings = embeddings.reshape(embeddings.shape[0], -1)
embeddings_sample = SimpleImputer(strategy="mean").fit_transform(embeddings[:sample_size])
pca_embeddings = PCA(n_components=50).fit_transform(embeddings_sample)
tsne_embeddings = TSNE(n_components=2, perplexity=30, random_state=42).fit_transform(pca_embeddings)
adjusted_clusters = min(optimal_clusters, len(tsne_embeddings), 10)
adjusted_clusters = max(2, adjusted_clusters)
kmeans_emb = KMeans(n_clusters=adjusted_clusters, random_state=42, n_init=10).fit(tsne_embeddings)
dist_emb = squareform(pdist(kmeans_emb.cluster_centers_, metric='euclidean'))
min_size = min(dist_orig.size, dist_emb.size)
dist_orig_resized = dist_orig.flatten()[:min_size]
dist_emb_resized = dist_emb.flatten()[:min_size]
correlation, _ = spearmanr(dist_orig_resized, dist_emb_resized)
cosine_sim = cosine_similarity(dist_orig_resized.reshape(1, -1), dist_emb_resized.reshape(1, -1))[0][0]
results.append({
"Dataset": dataset_name,
"Function Type": dataset_name.split("_")[-1],
"Model": model_name,
"Optimal Clusters": optimal_clusters,
"Adjusted Clusters (Embeddings)": adjusted_clusters,
"Distance Correlation": round(correlation, 3),
"Cosine Similarity": round(cosine_sim, 3)
})
return results
# ✅ RUN PIPELINE
all_results = []
for set_info in processing_sets:
dataframes, embeddings_data = load_data(set_info["input_dir"], set_info["output_dir"], set_info["file_ext"], set_info["models"])
if embeddings_data:
results = compute_cluster_similarity(dataframes, embeddings_data)
all_results.extend(results)
# ✅ Save to CSV
results_df = pd.DataFrame(all_results)
if not results_df.empty:
results_df.to_csv("cluster_similarity_results.csv", index=False)
print(f"✅ Results saved.")
📂 Processing /home/abbasali/synthetic data/generated_data -> /home/abbasali/synthetic data/embeddings ✅ Loaded 100 CSV datasets from /home/abbasali/synthetic data/generated_data. ✅ Loaded embeddings for data_79_quadratic (Model: bert) ✅ Loaded embeddings for data_79_quadratic (Model: ernie) ✅ Loaded embeddings for data_79_quadratic (Model: gatortron) ✅ Loaded embeddings for data_79_quadratic (Model: roberta) ✅ Loaded embeddings for data_79_quadratic (Model: t5) ✅ Loaded embeddings for data_79_quadratic (Model: e5_small) ✅ Loaded embeddings for data_79_quadratic (Model: minilm) ✅ Loaded embeddings for data_91_linear (Model: bert) ✅ Loaded embeddings for data_91_linear (Model: ernie) ✅ Loaded embeddings for data_91_linear (Model: gatortron) ✅ Loaded embeddings for data_91_linear (Model: roberta) ✅ Loaded embeddings for data_91_linear (Model: t5) ✅ Loaded embeddings for data_91_linear (Model: e5_small) ✅ Loaded embeddings for data_91_linear (Model: minilm) ✅ Loaded embeddings for data_44_exponential (Model: bert) ✅ Loaded embeddings for data_44_exponential (Model: ernie) ✅ Loaded embeddings for data_44_exponential (Model: gatortron) ✅ Loaded embeddings for data_44_exponential (Model: roberta) ✅ Loaded embeddings for data_44_exponential (Model: t5) ✅ Loaded embeddings for data_44_exponential (Model: e5_small) ✅ Loaded embeddings for data_44_exponential (Model: minilm) ✅ Loaded embeddings for data_42_linear (Model: bert) ✅ Loaded embeddings for data_42_linear (Model: ernie) ✅ Loaded embeddings for data_42_linear (Model: gatortron) ✅ Loaded embeddings for data_42_linear (Model: roberta) ✅ Loaded embeddings for data_42_linear (Model: t5) ✅ Loaded embeddings for data_42_linear (Model: e5_small) ✅ Loaded embeddings for data_42_linear (Model: minilm) ✅ Loaded embeddings for data_88_exponential (Model: bert) ✅ Loaded embeddings for data_88_exponential (Model: ernie) ✅ Loaded embeddings for data_88_exponential (Model: gatortron) ✅ Loaded embeddings for data_88_exponential (Model: roberta) ✅ Loaded embeddings for data_88_exponential (Model: t5) ✅ Loaded embeddings for data_88_exponential (Model: e5_small) ✅ Loaded embeddings for data_88_exponential (Model: minilm) ✅ Loaded embeddings for data_43_linear (Model: bert) ✅ Loaded embeddings for data_43_linear (Model: ernie) ✅ Loaded embeddings for data_43_linear (Model: gatortron) ✅ Loaded embeddings for data_43_linear (Model: roberta) ✅ Loaded embeddings for data_43_linear (Model: t5) ✅ Loaded embeddings for data_43_linear (Model: e5_small) ✅ Loaded embeddings for data_43_linear (Model: minilm) ✅ Loaded embeddings for data_68_exponential (Model: bert) ✅ Loaded embeddings for data_68_exponential (Model: ernie) ✅ Loaded embeddings for data_68_exponential (Model: gatortron) ✅ Loaded embeddings for data_68_exponential (Model: roberta) ✅ Loaded embeddings for data_68_exponential (Model: t5) ✅ Loaded embeddings for data_68_exponential (Model: e5_small) ✅ Loaded embeddings for data_68_exponential (Model: minilm) ✅ Loaded embeddings for data_48_cubic (Model: bert) ✅ Loaded embeddings for data_48_cubic (Model: ernie) ✅ Loaded embeddings for data_48_cubic (Model: gatortron) ✅ Loaded embeddings for data_48_cubic (Model: roberta) ✅ Loaded embeddings for data_48_cubic (Model: t5) ✅ Loaded embeddings for data_48_cubic (Model: e5_small) ✅ Loaded embeddings for data_48_cubic (Model: minilm) ✅ Loaded embeddings for data_13_cubic (Model: bert) ✅ Loaded embeddings for data_13_cubic (Model: ernie) ✅ Loaded embeddings for data_13_cubic (Model: gatortron) ✅ Loaded embeddings for data_13_cubic (Model: roberta) ✅ Loaded embeddings for data_13_cubic (Model: t5) ✅ Loaded embeddings for data_13_cubic (Model: e5_small) ✅ Loaded embeddings for data_13_cubic (Model: minilm) ✅ Loaded embeddings for data_33_quadratic (Model: bert) ✅ Loaded embeddings for data_33_quadratic (Model: ernie) ✅ Loaded embeddings for data_33_quadratic (Model: gatortron) ✅ Loaded embeddings for data_33_quadratic (Model: roberta) ✅ Loaded embeddings for data_33_quadratic (Model: t5) ✅ Loaded embeddings for data_33_quadratic (Model: e5_small) ✅ Loaded embeddings for data_33_quadratic (Model: minilm) ✅ Loaded embeddings for data_84_cubic (Model: bert) ✅ Loaded embeddings for data_84_cubic (Model: ernie) ✅ Loaded embeddings for data_84_cubic (Model: gatortron) ✅ Loaded embeddings for data_84_cubic (Model: roberta) ✅ Loaded embeddings for data_84_cubic (Model: t5) ✅ Loaded embeddings for data_84_cubic (Model: e5_small) ✅ Loaded embeddings for data_84_cubic (Model: minilm) ✅ Loaded embeddings for data_74_linear (Model: bert) ✅ Loaded embeddings for data_74_linear (Model: ernie) ✅ Loaded embeddings for data_74_linear (Model: gatortron) ✅ Loaded embeddings for data_74_linear (Model: roberta) ✅ Loaded embeddings for data_74_linear (Model: t5) ✅ Loaded embeddings for data_74_linear (Model: e5_small) ✅ Loaded embeddings for data_74_linear (Model: minilm) ✅ Loaded embeddings for data_21_quadratic (Model: bert) ✅ Loaded embeddings for data_21_quadratic (Model: ernie) ✅ Loaded embeddings for data_21_quadratic (Model: gatortron) ✅ Loaded embeddings for data_21_quadratic (Model: roberta) ✅ Loaded embeddings for data_21_quadratic (Model: t5) ✅ Loaded embeddings for data_21_quadratic (Model: e5_small) ✅ Loaded embeddings for data_21_quadratic (Model: minilm) ✅ Loaded embeddings for data_40_exponential (Model: bert) ✅ Loaded embeddings for data_40_exponential (Model: ernie) ✅ Loaded embeddings for data_40_exponential (Model: gatortron) ✅ Loaded embeddings for data_40_exponential (Model: roberta) ✅ Loaded embeddings for data_40_exponential (Model: t5) ✅ Loaded embeddings for data_40_exponential (Model: e5_small) ✅ Loaded embeddings for data_40_exponential (Model: minilm) ✅ Loaded embeddings for data_5_cubic (Model: bert) ✅ Loaded embeddings for data_5_cubic (Model: ernie) ✅ Loaded embeddings for data_5_cubic (Model: gatortron) ✅ Loaded embeddings for data_5_cubic (Model: roberta) ✅ Loaded embeddings for data_5_cubic (Model: t5) ✅ Loaded embeddings for data_5_cubic (Model: e5_small) ✅ Loaded embeddings for data_5_cubic (Model: minilm) ✅ Loaded embeddings for data_17_exponential (Model: bert) ✅ Loaded embeddings for data_17_exponential (Model: ernie) ✅ Loaded embeddings for data_17_exponential (Model: gatortron) ✅ Loaded embeddings for data_17_exponential (Model: roberta) ✅ Loaded embeddings for data_17_exponential (Model: t5) ✅ Loaded embeddings for data_17_exponential (Model: e5_small) ✅ Loaded embeddings for data_17_exponential (Model: minilm) ✅ Loaded embeddings for data_0_exponential (Model: bert) ✅ Loaded embeddings for data_0_exponential (Model: ernie) ✅ Loaded embeddings for data_0_exponential (Model: gatortron) ✅ Loaded embeddings for data_0_exponential (Model: roberta) ✅ Loaded embeddings for data_0_exponential (Model: t5) ✅ Loaded embeddings for data_0_exponential (Model: e5_small) ✅ Loaded embeddings for data_0_exponential (Model: minilm) ✅ Loaded embeddings for data_30_cubic (Model: bert) ✅ Loaded embeddings for data_30_cubic (Model: ernie) ✅ Loaded embeddings for data_30_cubic (Model: gatortron) ✅ Loaded embeddings for data_30_cubic (Model: roberta) ✅ Loaded embeddings for data_30_cubic (Model: t5) ✅ Loaded embeddings for data_30_cubic (Model: e5_small) ✅ Loaded embeddings for data_30_cubic (Model: minilm) ✅ Loaded embeddings for data_34_cubic (Model: bert) ✅ Loaded embeddings for data_34_cubic (Model: ernie) ✅ Loaded embeddings for data_34_cubic (Model: gatortron) ✅ Loaded embeddings for data_34_cubic (Model: roberta) ✅ Loaded embeddings for data_34_cubic (Model: t5) ✅ Loaded embeddings for data_34_cubic (Model: e5_small) ✅ Loaded embeddings for data_34_cubic (Model: minilm) ✅ Loaded embeddings for data_52_cubic (Model: bert) ✅ Loaded embeddings for data_52_cubic (Model: ernie) ✅ Loaded embeddings for data_52_cubic (Model: gatortron) ✅ Loaded embeddings for data_52_cubic (Model: roberta) ✅ Loaded embeddings for data_52_cubic (Model: t5) ✅ Loaded embeddings for data_52_cubic (Model: e5_small) ✅ Loaded embeddings for data_52_cubic (Model: minilm) ✅ Loaded embeddings for data_16_cubic (Model: bert) ✅ Loaded embeddings for data_16_cubic (Model: ernie) ✅ Loaded embeddings for data_16_cubic (Model: gatortron) ✅ Loaded embeddings for data_16_cubic (Model: roberta) ✅ Loaded embeddings for data_16_cubic (Model: t5) ✅ Loaded embeddings for data_16_cubic (Model: e5_small) ✅ Loaded embeddings for data_16_cubic (Model: minilm) ✅ Loaded embeddings for data_72_linear (Model: bert) ✅ Loaded embeddings for data_72_linear (Model: ernie) ✅ Loaded embeddings for data_72_linear (Model: gatortron) ✅ Loaded embeddings for data_72_linear (Model: roberta) ✅ Loaded embeddings for data_72_linear (Model: t5) ✅ Loaded embeddings for data_72_linear (Model: e5_small) ✅ Loaded embeddings for data_72_linear (Model: minilm) ✅ Loaded embeddings for data_89_linear (Model: bert) ✅ Loaded embeddings for data_89_linear (Model: ernie) ✅ Loaded embeddings for data_89_linear (Model: gatortron) ✅ Loaded embeddings for data_89_linear (Model: roberta) ✅ Loaded embeddings for data_89_linear (Model: t5) ✅ Loaded embeddings for data_89_linear (Model: e5_small) ✅ Loaded embeddings for data_89_linear (Model: minilm) ✅ Loaded embeddings for data_55_linear (Model: bert) ✅ Loaded embeddings for data_55_linear (Model: ernie) ✅ Loaded embeddings for data_55_linear (Model: gatortron) ✅ Loaded embeddings for data_55_linear (Model: roberta) ✅ Loaded embeddings for data_55_linear (Model: t5) ✅ Loaded embeddings for data_55_linear (Model: e5_small) ✅ Loaded embeddings for data_55_linear (Model: minilm) ✅ Loaded embeddings for data_22_quadratic (Model: bert) ✅ Loaded embeddings for data_22_quadratic (Model: ernie) ✅ Loaded embeddings for data_22_quadratic (Model: gatortron) ✅ Loaded embeddings for data_22_quadratic (Model: roberta) ✅ Loaded embeddings for data_22_quadratic (Model: t5) ✅ Loaded embeddings for data_22_quadratic (Model: e5_small) ✅ Loaded embeddings for data_22_quadratic (Model: minilm) ✅ Loaded embeddings for data_65_exponential (Model: bert) ✅ Loaded embeddings for data_65_exponential (Model: ernie) ✅ Loaded embeddings for data_65_exponential (Model: gatortron) ✅ Loaded embeddings for data_65_exponential (Model: roberta) ✅ Loaded embeddings for data_65_exponential (Model: t5) ✅ Loaded embeddings for data_65_exponential (Model: e5_small) ✅ Loaded embeddings for data_65_exponential (Model: minilm) ✅ Loaded embeddings for data_14_linear (Model: bert) ✅ Loaded embeddings for data_14_linear (Model: ernie) ✅ Loaded embeddings for data_14_linear (Model: gatortron) ✅ Loaded embeddings for data_14_linear (Model: roberta) ✅ Loaded embeddings for data_14_linear (Model: t5) ✅ Loaded embeddings for data_14_linear (Model: e5_small) ✅ Loaded embeddings for data_14_linear (Model: minilm) ✅ Loaded embeddings for data_25_exponential (Model: bert) ✅ Loaded embeddings for data_25_exponential (Model: ernie) ✅ Loaded embeddings for data_25_exponential (Model: gatortron) ✅ Loaded embeddings for data_25_exponential (Model: roberta) ✅ Loaded embeddings for data_25_exponential (Model: t5) ✅ Loaded embeddings for data_25_exponential (Model: e5_small) ✅ Loaded embeddings for data_25_exponential (Model: minilm) ✅ Loaded embeddings for data_94_quadratic (Model: bert) ✅ Loaded embeddings for data_94_quadratic (Model: ernie) ✅ Loaded embeddings for data_94_quadratic (Model: gatortron) ✅ Loaded embeddings for data_94_quadratic (Model: roberta) ✅ Loaded embeddings for data_94_quadratic (Model: t5) ✅ Loaded embeddings for data_94_quadratic (Model: e5_small) ✅ Loaded embeddings for data_94_quadratic (Model: minilm) ✅ Loaded embeddings for data_96_cubic (Model: bert) ✅ Loaded embeddings for data_96_cubic (Model: ernie) ✅ Loaded embeddings for data_96_cubic (Model: gatortron) ✅ Loaded embeddings for data_96_cubic (Model: roberta) ✅ Loaded embeddings for data_96_cubic (Model: t5) ✅ Loaded embeddings for data_96_cubic (Model: e5_small) ✅ Loaded embeddings for data_96_cubic (Model: minilm) ✅ Loaded embeddings for data_85_quadratic (Model: bert) ✅ Loaded embeddings for data_85_quadratic (Model: ernie) ✅ Loaded embeddings for data_85_quadratic (Model: gatortron) ✅ Loaded embeddings for data_85_quadratic (Model: roberta) ✅ Loaded embeddings for data_85_quadratic (Model: t5) ✅ Loaded embeddings for data_85_quadratic (Model: e5_small) ✅ Loaded embeddings for data_85_quadratic (Model: minilm) ✅ Loaded embeddings for data_51_quadratic (Model: bert) ✅ Loaded embeddings for data_51_quadratic (Model: ernie) ✅ Loaded embeddings for data_51_quadratic (Model: gatortron) ✅ Loaded embeddings for data_51_quadratic (Model: roberta) ✅ Loaded embeddings for data_51_quadratic (Model: t5) ✅ Loaded embeddings for data_51_quadratic (Model: e5_small) ✅ Loaded embeddings for data_51_quadratic (Model: minilm) ✅ Loaded embeddings for data_90_quadratic (Model: bert) ✅ Loaded embeddings for data_90_quadratic (Model: ernie) ✅ Loaded embeddings for data_90_quadratic (Model: gatortron) ✅ Loaded embeddings for data_90_quadratic (Model: roberta) ✅ Loaded embeddings for data_90_quadratic (Model: t5) ✅ Loaded embeddings for data_90_quadratic (Model: e5_small) ✅ Loaded embeddings for data_90_quadratic (Model: minilm) ✅ Loaded embeddings for data_38_exponential (Model: bert) ✅ Loaded embeddings for data_38_exponential (Model: ernie) ✅ Loaded embeddings for data_38_exponential (Model: gatortron) ✅ Loaded embeddings for data_38_exponential (Model: roberta) ✅ Loaded embeddings for data_38_exponential (Model: t5) ✅ Loaded embeddings for data_38_exponential (Model: e5_small) ✅ Loaded embeddings for data_38_exponential (Model: minilm) ✅ Loaded embeddings for data_81_exponential (Model: bert) ✅ Loaded embeddings for data_81_exponential (Model: ernie) ✅ Loaded embeddings for data_81_exponential (Model: gatortron) ✅ Loaded embeddings for data_81_exponential (Model: roberta) ✅ Loaded embeddings for data_81_exponential (Model: t5) ✅ Loaded embeddings for data_81_exponential (Model: e5_small) ✅ Loaded embeddings for data_81_exponential (Model: minilm) ✅ Loaded embeddings for data_92_exponential (Model: bert) ✅ Loaded embeddings for data_92_exponential (Model: ernie) ✅ Loaded embeddings for data_92_exponential (Model: gatortron) ✅ Loaded embeddings for data_92_exponential (Model: roberta) ✅ Loaded embeddings for data_92_exponential (Model: t5) ✅ Loaded embeddings for data_92_exponential (Model: e5_small) ✅ Loaded embeddings for data_92_exponential (Model: minilm) ✅ Loaded embeddings for data_62_quadratic (Model: bert) ✅ Loaded embeddings for data_62_quadratic (Model: ernie) ✅ Loaded embeddings for data_62_quadratic (Model: gatortron) ✅ Loaded embeddings for data_62_quadratic (Model: roberta) ✅ Loaded embeddings for data_62_quadratic (Model: t5) ✅ Loaded embeddings for data_62_quadratic (Model: e5_small) ✅ Loaded embeddings for data_62_quadratic (Model: minilm) ✅ Loaded embeddings for data_10_linear (Model: bert) ✅ Loaded embeddings for data_10_linear (Model: ernie) ✅ Loaded embeddings for data_10_linear (Model: gatortron) ✅ Loaded embeddings for data_10_linear (Model: roberta) ✅ Loaded embeddings for data_10_linear (Model: t5) ✅ Loaded embeddings for data_10_linear (Model: e5_small) ✅ Loaded embeddings for data_10_linear (Model: minilm) ✅ Loaded embeddings for data_31_linear (Model: bert) ✅ Loaded embeddings for data_31_linear (Model: ernie) ✅ Loaded embeddings for data_31_linear (Model: gatortron) ✅ Loaded embeddings for data_31_linear (Model: roberta) ✅ Loaded embeddings for data_31_linear (Model: t5) ✅ Loaded embeddings for data_31_linear (Model: e5_small) ✅ Loaded embeddings for data_31_linear (Model: minilm) ✅ Loaded embeddings for data_57_linear (Model: bert) ✅ Loaded embeddings for data_57_linear (Model: ernie) ✅ Loaded embeddings for data_57_linear (Model: gatortron) ✅ Loaded embeddings for data_57_linear (Model: roberta) ✅ Loaded embeddings for data_57_linear (Model: t5) ✅ Loaded embeddings for data_57_linear (Model: e5_small) ✅ Loaded embeddings for data_57_linear (Model: minilm) ✅ Loaded embeddings for data_70_linear (Model: bert) ✅ Loaded embeddings for data_70_linear (Model: ernie) ✅ Loaded embeddings for data_70_linear (Model: gatortron) ✅ Loaded embeddings for data_70_linear (Model: roberta) ✅ Loaded embeddings for data_70_linear (Model: t5) ✅ Loaded embeddings for data_70_linear (Model: e5_small) ✅ Loaded embeddings for data_70_linear (Model: minilm) ✅ Loaded embeddings for data_98_cubic (Model: bert) ✅ Loaded embeddings for data_98_cubic (Model: ernie) ✅ Loaded embeddings for data_98_cubic (Model: gatortron) ✅ Loaded embeddings for data_98_cubic (Model: roberta) ✅ Loaded embeddings for data_98_cubic (Model: t5) ✅ Loaded embeddings for data_98_cubic (Model: e5_small) ✅ Loaded embeddings for data_98_cubic (Model: minilm) ✅ Loaded embeddings for data_23_cubic (Model: bert) ✅ Loaded embeddings for data_23_cubic (Model: ernie) ✅ Loaded embeddings for data_23_cubic (Model: gatortron) ✅ Loaded embeddings for data_23_cubic (Model: roberta) ✅ Loaded embeddings for data_23_cubic (Model: t5) ✅ Loaded embeddings for data_23_cubic (Model: e5_small) ✅ Loaded embeddings for data_23_cubic (Model: minilm) ✅ Loaded embeddings for data_27_exponential (Model: bert) ✅ Loaded embeddings for data_27_exponential (Model: ernie) ✅ Loaded embeddings for data_27_exponential (Model: gatortron) ✅ Loaded embeddings for data_27_exponential (Model: roberta) ✅ Loaded embeddings for data_27_exponential (Model: t5) ✅ Loaded embeddings for data_27_exponential (Model: e5_small) ✅ Loaded embeddings for data_27_exponential (Model: minilm) ✅ Loaded embeddings for data_61_exponential (Model: bert) ✅ Loaded embeddings for data_61_exponential (Model: ernie) ✅ Loaded embeddings for data_61_exponential (Model: gatortron) ✅ Loaded embeddings for data_61_exponential (Model: roberta) ✅ Loaded embeddings for data_61_exponential (Model: t5) ✅ Loaded embeddings for data_61_exponential (Model: e5_small) ✅ Loaded embeddings for data_61_exponential (Model: minilm) ✅ Loaded embeddings for data_46_linear (Model: bert) ✅ Loaded embeddings for data_46_linear (Model: ernie) ✅ Loaded embeddings for data_46_linear (Model: gatortron) ✅ Loaded embeddings for data_46_linear (Model: roberta) ✅ Loaded embeddings for data_46_linear (Model: t5) ✅ Loaded embeddings for data_46_linear (Model: e5_small) ✅ Loaded embeddings for data_46_linear (Model: minilm) ✅ Loaded embeddings for data_73_cubic (Model: bert) ✅ Loaded embeddings for data_73_cubic (Model: ernie) ✅ Loaded embeddings for data_73_cubic (Model: gatortron) ✅ Loaded embeddings for data_73_cubic (Model: roberta) ✅ Loaded embeddings for data_73_cubic (Model: t5) ✅ Loaded embeddings for data_73_cubic (Model: e5_small) ✅ Loaded embeddings for data_73_cubic (Model: minilm) ✅ Loaded embeddings for data_53_quadratic (Model: bert) ✅ Loaded embeddings for data_53_quadratic (Model: ernie) ✅ Loaded embeddings for data_53_quadratic (Model: gatortron) ✅ Loaded embeddings for data_53_quadratic (Model: roberta) ✅ Loaded embeddings for data_53_quadratic (Model: t5) ✅ Loaded embeddings for data_53_quadratic (Model: e5_small) ✅ Loaded embeddings for data_53_quadratic (Model: minilm) ✅ Loaded embeddings for data_37_linear (Model: bert) ✅ Loaded embeddings for data_37_linear (Model: ernie) ✅ Loaded embeddings for data_37_linear (Model: gatortron) ✅ Loaded embeddings for data_37_linear (Model: roberta) ✅ Loaded embeddings for data_37_linear (Model: t5) ✅ Loaded embeddings for data_37_linear (Model: e5_small) ✅ Loaded embeddings for data_37_linear (Model: minilm) ✅ Loaded embeddings for data_54_quadratic (Model: bert) ✅ Loaded embeddings for data_54_quadratic (Model: ernie) ✅ Loaded embeddings for data_54_quadratic (Model: gatortron) ✅ Loaded embeddings for data_54_quadratic (Model: roberta) ✅ Loaded embeddings for data_54_quadratic (Model: t5) ✅ Loaded embeddings for data_54_quadratic (Model: e5_small) ✅ Loaded embeddings for data_54_quadratic (Model: minilm) ✅ Loaded embeddings for data_20_quadratic (Model: bert) ✅ Loaded embeddings for data_20_quadratic (Model: ernie) ✅ Loaded embeddings for data_20_quadratic (Model: gatortron) ✅ Loaded embeddings for data_20_quadratic (Model: roberta) ✅ Loaded embeddings for data_20_quadratic (Model: t5) ✅ Loaded embeddings for data_20_quadratic (Model: e5_small) ✅ Loaded embeddings for data_20_quadratic (Model: minilm) ✅ Loaded embeddings for data_78_quadratic (Model: bert) ✅ Loaded embeddings for data_78_quadratic (Model: ernie) ✅ Loaded embeddings for data_78_quadratic (Model: gatortron) ✅ Loaded embeddings for data_78_quadratic (Model: roberta) ✅ Loaded embeddings for data_78_quadratic (Model: t5) ✅ Loaded embeddings for data_78_quadratic (Model: e5_small) ✅ Loaded embeddings for data_78_quadratic (Model: minilm) ✅ Loaded embeddings for data_82_quadratic (Model: bert) ✅ Loaded embeddings for data_82_quadratic (Model: ernie) ✅ Loaded embeddings for data_82_quadratic (Model: gatortron) ✅ Loaded embeddings for data_82_quadratic (Model: roberta) ✅ Loaded embeddings for data_82_quadratic (Model: t5) ✅ Loaded embeddings for data_82_quadratic (Model: e5_small) ✅ Loaded embeddings for data_82_quadratic (Model: minilm) ✅ Loaded embeddings for data_19_linear (Model: bert) ✅ Loaded embeddings for data_19_linear (Model: ernie) ✅ Loaded embeddings for data_19_linear (Model: gatortron) ✅ Loaded embeddings for data_19_linear (Model: roberta) ✅ Loaded embeddings for data_19_linear (Model: t5) ✅ Loaded embeddings for data_19_linear (Model: e5_small) ✅ Loaded embeddings for data_19_linear (Model: minilm) ✅ Loaded embeddings for data_4_quadratic (Model: bert) ✅ Loaded embeddings for data_4_quadratic (Model: ernie) ✅ Loaded embeddings for data_4_quadratic (Model: gatortron) ✅ Loaded embeddings for data_4_quadratic (Model: roberta) ✅ Loaded embeddings for data_4_quadratic (Model: t5) ✅ Loaded embeddings for data_4_quadratic (Model: e5_small) ✅ Loaded embeddings for data_4_quadratic (Model: minilm) ✅ Loaded embeddings for data_8_quadratic (Model: bert) ✅ Loaded embeddings for data_8_quadratic (Model: ernie) ✅ Loaded embeddings for data_8_quadratic (Model: gatortron) ✅ Loaded embeddings for data_8_quadratic (Model: roberta) ✅ Loaded embeddings for data_8_quadratic (Model: t5) ✅ Loaded embeddings for data_8_quadratic (Model: e5_small) ✅ Loaded embeddings for data_8_quadratic (Model: minilm) ✅ Loaded embeddings for data_36_cubic (Model: bert) ✅ Loaded embeddings for data_36_cubic (Model: ernie) ✅ Loaded embeddings for data_36_cubic (Model: gatortron) ✅ Loaded embeddings for data_36_cubic (Model: roberta) ✅ Loaded embeddings for data_36_cubic (Model: t5) ✅ Loaded embeddings for data_36_cubic (Model: e5_small) ✅ Loaded embeddings for data_36_cubic (Model: minilm) ✅ Loaded embeddings for data_24_exponential (Model: bert) ✅ Loaded embeddings for data_24_exponential (Model: ernie) ✅ Loaded embeddings for data_24_exponential (Model: gatortron) ✅ Loaded embeddings for data_24_exponential (Model: roberta) ✅ Loaded embeddings for data_24_exponential (Model: t5) ✅ Loaded embeddings for data_24_exponential (Model: e5_small) ✅ Loaded embeddings for data_24_exponential (Model: minilm) ✅ Loaded embeddings for data_39_quadratic (Model: bert) ✅ Loaded embeddings for data_39_quadratic (Model: ernie) ✅ Loaded embeddings for data_39_quadratic (Model: gatortron) ✅ Loaded embeddings for data_39_quadratic (Model: roberta) ✅ Loaded embeddings for data_39_quadratic (Model: t5) ✅ Loaded embeddings for data_39_quadratic (Model: e5_small) ✅ Loaded embeddings for data_39_quadratic (Model: minilm) ✅ Loaded embeddings for data_60_cubic (Model: bert) ✅ Loaded embeddings for data_60_cubic (Model: ernie) ✅ Loaded embeddings for data_60_cubic (Model: gatortron) ✅ Loaded embeddings for data_60_cubic (Model: roberta) ✅ Loaded embeddings for data_60_cubic (Model: t5) ✅ Loaded embeddings for data_60_cubic (Model: e5_small) ✅ Loaded embeddings for data_60_cubic (Model: minilm) ✅ Loaded embeddings for data_1_quadratic (Model: bert) ✅ Loaded embeddings for data_1_quadratic (Model: ernie) ✅ Loaded embeddings for data_1_quadratic (Model: gatortron) ✅ Loaded embeddings for data_1_quadratic (Model: roberta) ✅ Loaded embeddings for data_1_quadratic (Model: t5) ✅ Loaded embeddings for data_1_quadratic (Model: e5_small) ✅ Loaded embeddings for data_1_quadratic (Model: minilm) ✅ Loaded embeddings for data_41_exponential (Model: bert) ✅ Loaded embeddings for data_41_exponential (Model: ernie) ✅ Loaded embeddings for data_41_exponential (Model: gatortron) ✅ Loaded embeddings for data_41_exponential (Model: roberta) ✅ Loaded embeddings for data_41_exponential (Model: t5) ✅ Loaded embeddings for data_41_exponential (Model: e5_small) ✅ Loaded embeddings for data_41_exponential (Model: minilm) ✅ Loaded embeddings for data_32_linear (Model: bert) ✅ Loaded embeddings for data_32_linear (Model: ernie) ✅ Loaded embeddings for data_32_linear (Model: gatortron) ✅ Loaded embeddings for data_32_linear (Model: roberta) ✅ Loaded embeddings for data_32_linear (Model: t5) ✅ Loaded embeddings for data_32_linear (Model: e5_small) ✅ Loaded embeddings for data_32_linear (Model: minilm) ✅ Loaded embeddings for data_66_linear (Model: bert) ✅ Loaded embeddings for data_66_linear (Model: ernie) ✅ Loaded embeddings for data_66_linear (Model: gatortron) ✅ Loaded embeddings for data_66_linear (Model: roberta) ✅ Loaded embeddings for data_66_linear (Model: t5) ✅ Loaded embeddings for data_66_linear (Model: e5_small) ✅ Loaded embeddings for data_66_linear (Model: minilm) ✅ Loaded embeddings for data_49_linear (Model: bert) ✅ Loaded embeddings for data_49_linear (Model: ernie) ✅ Loaded embeddings for data_49_linear (Model: gatortron) ✅ Loaded embeddings for data_49_linear (Model: roberta) ✅ Loaded embeddings for data_49_linear (Model: t5) ✅ Loaded embeddings for data_49_linear (Model: e5_small) ⚠️ WARNING: Missing embeddings for data_49_linear (Model: minilm) ✅ Loaded embeddings for data_56_linear (Model: bert) ✅ Loaded embeddings for data_56_linear (Model: ernie) ✅ Loaded embeddings for data_56_linear (Model: gatortron) ✅ Loaded embeddings for data_56_linear (Model: roberta) ✅ Loaded embeddings for data_56_linear (Model: t5) ✅ Loaded embeddings for data_56_linear (Model: e5_small) ✅ Loaded embeddings for data_56_linear (Model: minilm) ✅ Loaded embeddings for data_63_exponential (Model: bert) ✅ Loaded embeddings for data_63_exponential (Model: ernie) ✅ Loaded embeddings for data_63_exponential (Model: gatortron) ✅ Loaded embeddings for data_63_exponential (Model: roberta) ✅ Loaded embeddings for data_63_exponential (Model: t5) ✅ Loaded embeddings for data_63_exponential (Model: e5_small) ✅ Loaded embeddings for data_63_exponential (Model: minilm) ✅ Loaded embeddings for data_80_cubic (Model: bert) ✅ Loaded embeddings for data_80_cubic (Model: ernie) ✅ Loaded embeddings for data_80_cubic (Model: gatortron) ✅ Loaded embeddings for data_80_cubic (Model: roberta) ✅ Loaded embeddings for data_80_cubic (Model: t5) ✅ Loaded embeddings for data_80_cubic (Model: e5_small) ⚠️ WARNING: Missing embeddings for data_80_cubic (Model: minilm) ✅ Loaded embeddings for data_12_exponential (Model: bert) ✅ Loaded embeddings for data_12_exponential (Model: ernie) ✅ Loaded embeddings for data_12_exponential (Model: gatortron) ✅ Loaded embeddings for data_12_exponential (Model: roberta) ✅ Loaded embeddings for data_12_exponential (Model: t5) ✅ Loaded embeddings for data_12_exponential (Model: e5_small) ✅ Loaded embeddings for data_12_exponential (Model: minilm) ✅ Loaded embeddings for data_3_quadratic (Model: bert) ✅ Loaded embeddings for data_3_quadratic (Model: ernie) ✅ Loaded embeddings for data_3_quadratic (Model: gatortron) ✅ Loaded embeddings for data_3_quadratic (Model: roberta) ✅ Loaded embeddings for data_3_quadratic (Model: t5) ✅ Loaded embeddings for data_3_quadratic (Model: e5_small) ✅ Loaded embeddings for data_3_quadratic (Model: minilm) ✅ Loaded embeddings for data_64_exponential (Model: bert) ✅ Loaded embeddings for data_64_exponential (Model: ernie) ✅ Loaded embeddings for data_64_exponential (Model: gatortron) ✅ Loaded embeddings for data_64_exponential (Model: roberta) ✅ Loaded embeddings for data_64_exponential (Model: t5) ✅ Loaded embeddings for data_64_exponential (Model: e5_small) ✅ Loaded embeddings for data_64_exponential (Model: minilm) ✅ Loaded embeddings for data_35_cubic (Model: bert) ✅ Loaded embeddings for data_35_cubic (Model: ernie) ✅ Loaded embeddings for data_35_cubic (Model: gatortron) ✅ Loaded embeddings for data_35_cubic (Model: roberta) ✅ Loaded embeddings for data_35_cubic (Model: t5) ✅ Loaded embeddings for data_35_cubic (Model: e5_small) ✅ Loaded embeddings for data_35_cubic (Model: minilm) ✅ Loaded embeddings for data_6_linear (Model: bert) ✅ Loaded embeddings for data_6_linear (Model: ernie) ✅ Loaded embeddings for data_6_linear (Model: gatortron) ✅ Loaded embeddings for data_6_linear (Model: roberta) ✅ Loaded embeddings for data_6_linear (Model: t5) ✅ Loaded embeddings for data_6_linear (Model: e5_small) ✅ Loaded embeddings for data_6_linear (Model: minilm) ✅ Loaded embeddings for data_93_exponential (Model: bert) ✅ Loaded embeddings for data_93_exponential (Model: ernie) ✅ Loaded embeddings for data_93_exponential (Model: gatortron) ✅ Loaded embeddings for data_93_exponential (Model: roberta) ✅ Loaded embeddings for data_93_exponential (Model: t5) ✅ Loaded embeddings for data_93_exponential (Model: e5_small) ✅ Loaded embeddings for data_93_exponential (Model: minilm) ✅ Loaded embeddings for data_29_linear (Model: bert) ✅ Loaded embeddings for data_29_linear (Model: ernie) ✅ Loaded embeddings for data_29_linear (Model: gatortron) ✅ Loaded embeddings for data_29_linear (Model: roberta) ✅ Loaded embeddings for data_29_linear (Model: t5) ✅ Loaded embeddings for data_29_linear (Model: e5_small) ✅ Loaded embeddings for data_29_linear (Model: minilm) ✅ Loaded embeddings for data_76_cubic (Model: bert) ✅ Loaded embeddings for data_76_cubic (Model: ernie) ✅ Loaded embeddings for data_76_cubic (Model: gatortron) ✅ Loaded embeddings for data_76_cubic (Model: roberta) ✅ Loaded embeddings for data_76_cubic (Model: t5) ✅ Loaded embeddings for data_76_cubic (Model: e5_small) ✅ Loaded embeddings for data_76_cubic (Model: minilm) ✅ Loaded embeddings for data_50_exponential (Model: bert) ✅ Loaded embeddings for data_50_exponential (Model: ernie) ✅ Loaded embeddings for data_50_exponential (Model: gatortron) ✅ Loaded embeddings for data_50_exponential (Model: roberta) ✅ Loaded embeddings for data_50_exponential (Model: t5) ✅ Loaded embeddings for data_50_exponential (Model: e5_small) ✅ Loaded embeddings for data_50_exponential (Model: minilm) ✅ Loaded embeddings for data_9_exponential (Model: bert) ✅ Loaded embeddings for data_9_exponential (Model: ernie) ✅ Loaded embeddings for data_9_exponential (Model: gatortron) ✅ Loaded embeddings for data_9_exponential (Model: roberta) ✅ Loaded embeddings for data_9_exponential (Model: t5) ✅ Loaded embeddings for data_9_exponential (Model: e5_small) ✅ Loaded embeddings for data_9_exponential (Model: minilm) ✅ Loaded embeddings for data_59_quadratic (Model: bert) ✅ Loaded embeddings for data_59_quadratic (Model: ernie) ✅ Loaded embeddings for data_59_quadratic (Model: gatortron) ✅ Loaded embeddings for data_59_quadratic (Model: roberta) ✅ Loaded embeddings for data_59_quadratic (Model: t5) ✅ Loaded embeddings for data_59_quadratic (Model: e5_small) ✅ Loaded embeddings for data_59_quadratic (Model: minilm) ✅ Loaded embeddings for data_97_quadratic (Model: bert) ✅ Loaded embeddings for data_97_quadratic (Model: ernie) ✅ Loaded embeddings for data_97_quadratic (Model: gatortron) ✅ Loaded embeddings for data_97_quadratic (Model: roberta) ✅ Loaded embeddings for data_97_quadratic (Model: t5) ✅ Loaded embeddings for data_97_quadratic (Model: e5_small) ✅ Loaded embeddings for data_97_quadratic (Model: minilm) ✅ Loaded embeddings for data_45_cubic (Model: bert) ✅ Loaded embeddings for data_45_cubic (Model: ernie) ✅ Loaded embeddings for data_45_cubic (Model: gatortron) ✅ Loaded embeddings for data_45_cubic (Model: roberta) ✅ Loaded embeddings for data_45_cubic (Model: t5) ✅ Loaded embeddings for data_45_cubic (Model: e5_small) ✅ Loaded embeddings for data_45_cubic (Model: minilm) ✅ Loaded embeddings for data_18_quadratic (Model: bert) ✅ Loaded embeddings for data_18_quadratic (Model: ernie) ✅ Loaded embeddings for data_18_quadratic (Model: gatortron) ✅ Loaded embeddings for data_18_quadratic (Model: roberta) ✅ Loaded embeddings for data_18_quadratic (Model: t5) ✅ Loaded embeddings for data_18_quadratic (Model: e5_small) ✅ Loaded embeddings for data_18_quadratic (Model: minilm) ✅ Loaded embeddings for data_69_linear (Model: bert) ✅ Loaded embeddings for data_69_linear (Model: ernie) ✅ Loaded embeddings for data_69_linear (Model: gatortron) ✅ Loaded embeddings for data_69_linear (Model: roberta) ✅ Loaded embeddings for data_69_linear (Model: t5) ✅ Loaded embeddings for data_69_linear (Model: e5_small) ✅ Loaded embeddings for data_69_linear (Model: minilm) ✅ Loaded embeddings for data_67_linear (Model: bert) ✅ Loaded embeddings for data_67_linear (Model: ernie) ✅ Loaded embeddings for data_67_linear (Model: gatortron) ✅ Loaded embeddings for data_67_linear (Model: roberta) ✅ Loaded embeddings for data_67_linear (Model: t5) ✅ Loaded embeddings for data_67_linear (Model: e5_small) ✅ Loaded embeddings for data_67_linear (Model: minilm) ✅ Loaded embeddings for data_58_quadratic (Model: bert) ✅ Loaded embeddings for data_58_quadratic (Model: ernie) ✅ Loaded embeddings for data_58_quadratic (Model: gatortron) ✅ Loaded embeddings for data_58_quadratic (Model: roberta) ✅ Loaded embeddings for data_58_quadratic (Model: t5) ✅ Loaded embeddings for data_58_quadratic (Model: e5_small) ✅ Loaded embeddings for data_58_quadratic (Model: minilm) ✅ Loaded embeddings for data_95_quadratic (Model: bert) ✅ Loaded embeddings for data_95_quadratic (Model: ernie) ✅ Loaded embeddings for data_95_quadratic (Model: gatortron) ✅ Loaded embeddings for data_95_quadratic (Model: roberta) ✅ Loaded embeddings for data_95_quadratic (Model: t5) ✅ Loaded embeddings for data_95_quadratic (Model: e5_small) ✅ Loaded embeddings for data_95_quadratic (Model: minilm) ✅ Loaded embeddings for data_11_exponential (Model: bert) ✅ Loaded embeddings for data_11_exponential (Model: ernie) ✅ Loaded embeddings for data_11_exponential (Model: gatortron) ✅ Loaded embeddings for data_11_exponential (Model: roberta) ✅ Loaded embeddings for data_11_exponential (Model: t5) ✅ Loaded embeddings for data_11_exponential (Model: e5_small) ✅ Loaded embeddings for data_11_exponential (Model: minilm) ✅ Loaded embeddings for data_87_quadratic (Model: bert) ✅ Loaded embeddings for data_87_quadratic (Model: ernie) ✅ Loaded embeddings for data_87_quadratic (Model: gatortron) ✅ Loaded embeddings for data_87_quadratic (Model: roberta) ✅ Loaded embeddings for data_87_quadratic (Model: t5) ✅ Loaded embeddings for data_87_quadratic (Model: e5_small) ✅ Loaded embeddings for data_87_quadratic (Model: minilm) ✅ Loaded embeddings for data_2_linear (Model: bert) ✅ Loaded embeddings for data_2_linear (Model: ernie) ✅ Loaded embeddings for data_2_linear (Model: gatortron) ✅ Loaded embeddings for data_2_linear (Model: roberta) ✅ Loaded embeddings for data_2_linear (Model: t5) ✅ Loaded embeddings for data_2_linear (Model: e5_small) ✅ Loaded embeddings for data_2_linear (Model: minilm) ✅ Loaded embeddings for data_47_cubic (Model: bert) ✅ Loaded embeddings for data_47_cubic (Model: ernie) ✅ Loaded embeddings for data_47_cubic (Model: gatortron) ✅ Loaded embeddings for data_47_cubic (Model: roberta) ✅ Loaded embeddings for data_47_cubic (Model: t5) ✅ Loaded embeddings for data_47_cubic (Model: e5_small) ✅ Loaded embeddings for data_47_cubic (Model: minilm) ✅ Loaded embeddings for data_75_exponential (Model: bert) ✅ Loaded embeddings for data_75_exponential (Model: ernie) ✅ Loaded embeddings for data_75_exponential (Model: gatortron) ✅ Loaded embeddings for data_75_exponential (Model: roberta) ✅ Loaded embeddings for data_75_exponential (Model: t5) ✅ Loaded embeddings for data_75_exponential (Model: e5_small) ✅ Loaded embeddings for data_75_exponential (Model: minilm) ✅ Loaded embeddings for data_83_exponential (Model: bert) ✅ Loaded embeddings for data_83_exponential (Model: ernie) ✅ Loaded embeddings for data_83_exponential (Model: gatortron) ✅ Loaded embeddings for data_83_exponential (Model: roberta) ✅ Loaded embeddings for data_83_exponential (Model: t5) ✅ Loaded embeddings for data_83_exponential (Model: e5_small) ✅ Loaded embeddings for data_83_exponential (Model: minilm) ✅ Loaded embeddings for data_99_exponential (Model: bert) ✅ Loaded embeddings for data_99_exponential (Model: ernie) ✅ Loaded embeddings for data_99_exponential (Model: gatortron) ✅ Loaded embeddings for data_99_exponential (Model: roberta) ✅ Loaded embeddings for data_99_exponential (Model: t5) ✅ Loaded embeddings for data_99_exponential (Model: e5_small) ✅ Loaded embeddings for data_99_exponential (Model: minilm) ✅ Loaded embeddings for data_7_cubic (Model: bert) ✅ Loaded embeddings for data_7_cubic (Model: ernie) ✅ Loaded embeddings for data_7_cubic (Model: gatortron) ✅ Loaded embeddings for data_7_cubic (Model: roberta) ✅ Loaded embeddings for data_7_cubic (Model: t5) ✅ Loaded embeddings for data_7_cubic (Model: e5_small) ✅ Loaded embeddings for data_7_cubic (Model: minilm) ✅ Loaded embeddings for data_15_quadratic (Model: bert) ✅ Loaded embeddings for data_15_quadratic (Model: ernie) ✅ Loaded embeddings for data_15_quadratic (Model: gatortron) ✅ Loaded embeddings for data_15_quadratic (Model: roberta) ✅ Loaded embeddings for data_15_quadratic (Model: t5) ✅ Loaded embeddings for data_15_quadratic (Model: e5_small) ✅ Loaded embeddings for data_15_quadratic (Model: minilm) ✅ Loaded embeddings for data_71_cubic (Model: bert) ✅ Loaded embeddings for data_71_cubic (Model: ernie) ✅ Loaded embeddings for data_71_cubic (Model: gatortron) ✅ Loaded embeddings for data_71_cubic (Model: roberta) ✅ Loaded embeddings for data_71_cubic (Model: t5) ✅ Loaded embeddings for data_71_cubic (Model: e5_small) ✅ Loaded embeddings for data_71_cubic (Model: minilm) ✅ Loaded embeddings for data_77_cubic (Model: bert) ✅ Loaded embeddings for data_77_cubic (Model: ernie) ✅ Loaded embeddings for data_77_cubic (Model: gatortron) ✅ Loaded embeddings for data_77_cubic (Model: roberta) ✅ Loaded embeddings for data_77_cubic (Model: t5) ✅ Loaded embeddings for data_77_cubic (Model: e5_small) ✅ Loaded embeddings for data_77_cubic (Model: minilm) ✅ Loaded embeddings for data_86_linear (Model: bert) ✅ Loaded embeddings for data_86_linear (Model: ernie) ✅ Loaded embeddings for data_86_linear (Model: gatortron) ✅ Loaded embeddings for data_86_linear (Model: roberta) ✅ Loaded embeddings for data_86_linear (Model: t5) ✅ Loaded embeddings for data_86_linear (Model: e5_small) ✅ Loaded embeddings for data_86_linear (Model: minilm) ✅ Loaded embeddings for data_26_quadratic (Model: bert) ✅ Loaded embeddings for data_26_quadratic (Model: ernie) ✅ Loaded embeddings for data_26_quadratic (Model: gatortron) ✅ Loaded embeddings for data_26_quadratic (Model: roberta) ✅ Loaded embeddings for data_26_quadratic (Model: t5) ✅ Loaded embeddings for data_26_quadratic (Model: e5_small) ✅ Loaded embeddings for data_26_quadratic (Model: minilm) ✅ Loaded embeddings for data_28_linear (Model: bert) ✅ Loaded embeddings for data_28_linear (Model: ernie) ✅ Loaded embeddings for data_28_linear (Model: gatortron) ✅ Loaded embeddings for data_28_linear (Model: roberta) ✅ Loaded embeddings for data_28_linear (Model: t5) ✅ Loaded embeddings for data_28_linear (Model: e5_small) ✅ Loaded embeddings for data_28_linear (Model: minilm) 🔍 Computing cluster similarity for: data_79_quadratic 📊 Processing embeddings for data_79_quadratic (Model: bert) 📊 Processing embeddings for data_79_quadratic (Model: ernie) 📊 Processing embeddings for data_79_quadratic (Model: gatortron) 📊 Processing embeddings for data_79_quadratic (Model: roberta) 📊 Processing embeddings for data_79_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_79_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_79_quadratic (Model: e5_small) 📊 Processing embeddings for data_79_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_91_linear 📊 Processing embeddings for data_91_linear (Model: bert) 📊 Processing embeddings for data_91_linear (Model: ernie) 📊 Processing embeddings for data_91_linear (Model: gatortron) 📊 Processing embeddings for data_91_linear (Model: roberta) 📊 Processing embeddings for data_91_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_91_linear (Model: t5), flattening... 📊 Processing embeddings for data_91_linear (Model: e5_small) 📊 Processing embeddings for data_91_linear (Model: minilm) 🔍 Computing cluster similarity for: data_44_exponential 📊 Processing embeddings for data_44_exponential (Model: bert) 📊 Processing embeddings for data_44_exponential (Model: ernie) 📊 Processing embeddings for data_44_exponential (Model: gatortron) 📊 Processing embeddings for data_44_exponential (Model: roberta) 📊 Processing embeddings for data_44_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_44_exponential (Model: t5), flattening... 📊 Processing embeddings for data_44_exponential (Model: e5_small) 📊 Processing embeddings for data_44_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_42_linear 📊 Processing embeddings for data_42_linear (Model: bert) 📊 Processing embeddings for data_42_linear (Model: ernie) 📊 Processing embeddings for data_42_linear (Model: gatortron) 📊 Processing embeddings for data_42_linear (Model: roberta) 📊 Processing embeddings for data_42_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_42_linear (Model: t5), flattening... 📊 Processing embeddings for data_42_linear (Model: e5_small) 📊 Processing embeddings for data_42_linear (Model: minilm) 🔍 Computing cluster similarity for: data_88_exponential 📊 Processing embeddings for data_88_exponential (Model: bert) 📊 Processing embeddings for data_88_exponential (Model: ernie) 📊 Processing embeddings for data_88_exponential (Model: gatortron) 📊 Processing embeddings for data_88_exponential (Model: roberta) 📊 Processing embeddings for data_88_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_88_exponential (Model: t5), flattening... 📊 Processing embeddings for data_88_exponential (Model: e5_small) 📊 Processing embeddings for data_88_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_43_linear 📊 Processing embeddings for data_43_linear (Model: bert) 📊 Processing embeddings for data_43_linear (Model: ernie) 📊 Processing embeddings for data_43_linear (Model: gatortron) 📊 Processing embeddings for data_43_linear (Model: roberta) 📊 Processing embeddings for data_43_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_43_linear (Model: t5), flattening... 📊 Processing embeddings for data_43_linear (Model: e5_small) 📊 Processing embeddings for data_43_linear (Model: minilm) 🔍 Computing cluster similarity for: data_68_exponential 📊 Processing embeddings for data_68_exponential (Model: bert) 📊 Processing embeddings for data_68_exponential (Model: ernie) 📊 Processing embeddings for data_68_exponential (Model: gatortron) 📊 Processing embeddings for data_68_exponential (Model: roberta) 📊 Processing embeddings for data_68_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_68_exponential (Model: t5), flattening... 📊 Processing embeddings for data_68_exponential (Model: e5_small) 📊 Processing embeddings for data_68_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_48_cubic 📊 Processing embeddings for data_48_cubic (Model: bert) 📊 Processing embeddings for data_48_cubic (Model: ernie) 📊 Processing embeddings for data_48_cubic (Model: gatortron) 📊 Processing embeddings for data_48_cubic (Model: roberta) 📊 Processing embeddings for data_48_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_48_cubic (Model: t5), flattening... 📊 Processing embeddings for data_48_cubic (Model: e5_small) 📊 Processing embeddings for data_48_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_13_cubic 📊 Processing embeddings for data_13_cubic (Model: bert) 📊 Processing embeddings for data_13_cubic (Model: ernie) 📊 Processing embeddings for data_13_cubic (Model: gatortron) 📊 Processing embeddings for data_13_cubic (Model: roberta) 📊 Processing embeddings for data_13_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_13_cubic (Model: t5), flattening... 📊 Processing embeddings for data_13_cubic (Model: e5_small) 📊 Processing embeddings for data_13_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_33_quadratic 📊 Processing embeddings for data_33_quadratic (Model: bert) 📊 Processing embeddings for data_33_quadratic (Model: ernie) 📊 Processing embeddings for data_33_quadratic (Model: gatortron) 📊 Processing embeddings for data_33_quadratic (Model: roberta) 📊 Processing embeddings for data_33_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_33_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_33_quadratic (Model: e5_small) 📊 Processing embeddings for data_33_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_84_cubic 📊 Processing embeddings for data_84_cubic (Model: bert) 📊 Processing embeddings for data_84_cubic (Model: ernie) 📊 Processing embeddings for data_84_cubic (Model: gatortron) 📊 Processing embeddings for data_84_cubic (Model: roberta) 📊 Processing embeddings for data_84_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_84_cubic (Model: t5), flattening... 📊 Processing embeddings for data_84_cubic (Model: e5_small) 📊 Processing embeddings for data_84_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_74_linear 📊 Processing embeddings for data_74_linear (Model: bert) 📊 Processing embeddings for data_74_linear (Model: ernie) 📊 Processing embeddings for data_74_linear (Model: gatortron) 📊 Processing embeddings for data_74_linear (Model: roberta) 📊 Processing embeddings for data_74_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_74_linear (Model: t5), flattening... 📊 Processing embeddings for data_74_linear (Model: e5_small) 📊 Processing embeddings for data_74_linear (Model: minilm) 🔍 Computing cluster similarity for: data_21_quadratic 📊 Processing embeddings for data_21_quadratic (Model: bert) 📊 Processing embeddings for data_21_quadratic (Model: ernie) 📊 Processing embeddings for data_21_quadratic (Model: gatortron) 📊 Processing embeddings for data_21_quadratic (Model: roberta) 📊 Processing embeddings for data_21_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_21_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_21_quadratic (Model: e5_small) 📊 Processing embeddings for data_21_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_40_exponential 📊 Processing embeddings for data_40_exponential (Model: bert) 📊 Processing embeddings for data_40_exponential (Model: ernie) 📊 Processing embeddings for data_40_exponential (Model: gatortron) 📊 Processing embeddings for data_40_exponential (Model: roberta) 📊 Processing embeddings for data_40_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_40_exponential (Model: t5), flattening... 📊 Processing embeddings for data_40_exponential (Model: e5_small) 📊 Processing embeddings for data_40_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_5_cubic 📊 Processing embeddings for data_5_cubic (Model: bert) 📊 Processing embeddings for data_5_cubic (Model: ernie) 📊 Processing embeddings for data_5_cubic (Model: gatortron) 📊 Processing embeddings for data_5_cubic (Model: roberta) 📊 Processing embeddings for data_5_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_5_cubic (Model: t5), flattening... 📊 Processing embeddings for data_5_cubic (Model: e5_small) 📊 Processing embeddings for data_5_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_17_exponential 📊 Processing embeddings for data_17_exponential (Model: bert) 📊 Processing embeddings for data_17_exponential (Model: ernie) 📊 Processing embeddings for data_17_exponential (Model: gatortron) 📊 Processing embeddings for data_17_exponential (Model: roberta) 📊 Processing embeddings for data_17_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_17_exponential (Model: t5), flattening... 📊 Processing embeddings for data_17_exponential (Model: e5_small) 📊 Processing embeddings for data_17_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_0_exponential 📊 Processing embeddings for data_0_exponential (Model: bert) 📊 Processing embeddings for data_0_exponential (Model: ernie) 📊 Processing embeddings for data_0_exponential (Model: gatortron) 📊 Processing embeddings for data_0_exponential (Model: roberta) 📊 Processing embeddings for data_0_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_0_exponential (Model: t5), flattening... 📊 Processing embeddings for data_0_exponential (Model: e5_small) 📊 Processing embeddings for data_0_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_30_cubic 📊 Processing embeddings for data_30_cubic (Model: bert) 📊 Processing embeddings for data_30_cubic (Model: ernie) 📊 Processing embeddings for data_30_cubic (Model: gatortron) 📊 Processing embeddings for data_30_cubic (Model: roberta) 📊 Processing embeddings for data_30_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_30_cubic (Model: t5), flattening... 📊 Processing embeddings for data_30_cubic (Model: e5_small) 📊 Processing embeddings for data_30_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_34_cubic 📊 Processing embeddings for data_34_cubic (Model: bert) 📊 Processing embeddings for data_34_cubic (Model: ernie) 📊 Processing embeddings for data_34_cubic (Model: gatortron) 📊 Processing embeddings for data_34_cubic (Model: roberta) 📊 Processing embeddings for data_34_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_34_cubic (Model: t5), flattening... 📊 Processing embeddings for data_34_cubic (Model: e5_small) 📊 Processing embeddings for data_34_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_52_cubic 📊 Processing embeddings for data_52_cubic (Model: bert) 📊 Processing embeddings for data_52_cubic (Model: ernie) 📊 Processing embeddings for data_52_cubic (Model: gatortron) 📊 Processing embeddings for data_52_cubic (Model: roberta) 📊 Processing embeddings for data_52_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_52_cubic (Model: t5), flattening... 📊 Processing embeddings for data_52_cubic (Model: e5_small) 📊 Processing embeddings for data_52_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_16_cubic 📊 Processing embeddings for data_16_cubic (Model: bert) 📊 Processing embeddings for data_16_cubic (Model: ernie) 📊 Processing embeddings for data_16_cubic (Model: gatortron) 📊 Processing embeddings for data_16_cubic (Model: roberta) 📊 Processing embeddings for data_16_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_16_cubic (Model: t5), flattening... 📊 Processing embeddings for data_16_cubic (Model: e5_small) 📊 Processing embeddings for data_16_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_72_linear 📊 Processing embeddings for data_72_linear (Model: bert) 📊 Processing embeddings for data_72_linear (Model: ernie) 📊 Processing embeddings for data_72_linear (Model: gatortron) 📊 Processing embeddings for data_72_linear (Model: roberta) 📊 Processing embeddings for data_72_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_72_linear (Model: t5), flattening... 📊 Processing embeddings for data_72_linear (Model: e5_small) 📊 Processing embeddings for data_72_linear (Model: minilm) 🔍 Computing cluster similarity for: data_89_linear 📊 Processing embeddings for data_89_linear (Model: bert) 📊 Processing embeddings for data_89_linear (Model: ernie) 📊 Processing embeddings for data_89_linear (Model: gatortron) 📊 Processing embeddings for data_89_linear (Model: roberta) 📊 Processing embeddings for data_89_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_89_linear (Model: t5), flattening... 📊 Processing embeddings for data_89_linear (Model: e5_small) 📊 Processing embeddings for data_89_linear (Model: minilm) 🔍 Computing cluster similarity for: data_55_linear 📊 Processing embeddings for data_55_linear (Model: bert) 📊 Processing embeddings for data_55_linear (Model: ernie) 📊 Processing embeddings for data_55_linear (Model: gatortron) 📊 Processing embeddings for data_55_linear (Model: roberta) 📊 Processing embeddings for data_55_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_55_linear (Model: t5), flattening... 📊 Processing embeddings for data_55_linear (Model: e5_small) 📊 Processing embeddings for data_55_linear (Model: minilm) 🔍 Computing cluster similarity for: data_22_quadratic 📊 Processing embeddings for data_22_quadratic (Model: bert) 📊 Processing embeddings for data_22_quadratic (Model: ernie) 📊 Processing embeddings for data_22_quadratic (Model: gatortron) 📊 Processing embeddings for data_22_quadratic (Model: roberta) 📊 Processing embeddings for data_22_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_22_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_22_quadratic (Model: e5_small) 📊 Processing embeddings for data_22_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_65_exponential 📊 Processing embeddings for data_65_exponential (Model: bert) 📊 Processing embeddings for data_65_exponential (Model: ernie) 📊 Processing embeddings for data_65_exponential (Model: gatortron) 📊 Processing embeddings for data_65_exponential (Model: roberta) 📊 Processing embeddings for data_65_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_65_exponential (Model: t5), flattening... 📊 Processing embeddings for data_65_exponential (Model: e5_small) 📊 Processing embeddings for data_65_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_14_linear 📊 Processing embeddings for data_14_linear (Model: bert) 📊 Processing embeddings for data_14_linear (Model: ernie) 📊 Processing embeddings for data_14_linear (Model: gatortron) 📊 Processing embeddings for data_14_linear (Model: roberta) 📊 Processing embeddings for data_14_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_14_linear (Model: t5), flattening... 📊 Processing embeddings for data_14_linear (Model: e5_small) 📊 Processing embeddings for data_14_linear (Model: minilm) 🔍 Computing cluster similarity for: data_25_exponential 📊 Processing embeddings for data_25_exponential (Model: bert) 📊 Processing embeddings for data_25_exponential (Model: ernie) 📊 Processing embeddings for data_25_exponential (Model: gatortron) 📊 Processing embeddings for data_25_exponential (Model: roberta) 📊 Processing embeddings for data_25_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_25_exponential (Model: t5), flattening... 📊 Processing embeddings for data_25_exponential (Model: e5_small) 📊 Processing embeddings for data_25_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_94_quadratic 📊 Processing embeddings for data_94_quadratic (Model: bert) 📊 Processing embeddings for data_94_quadratic (Model: ernie) 📊 Processing embeddings for data_94_quadratic (Model: gatortron) 📊 Processing embeddings for data_94_quadratic (Model: roberta) 📊 Processing embeddings for data_94_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_94_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_94_quadratic (Model: e5_small) 📊 Processing embeddings for data_94_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_96_cubic 📊 Processing embeddings for data_96_cubic (Model: bert) 📊 Processing embeddings for data_96_cubic (Model: ernie) 📊 Processing embeddings for data_96_cubic (Model: gatortron) 📊 Processing embeddings for data_96_cubic (Model: roberta) 📊 Processing embeddings for data_96_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_96_cubic (Model: t5), flattening... 📊 Processing embeddings for data_96_cubic (Model: e5_small) 📊 Processing embeddings for data_96_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_85_quadratic 📊 Processing embeddings for data_85_quadratic (Model: bert) 📊 Processing embeddings for data_85_quadratic (Model: ernie) 📊 Processing embeddings for data_85_quadratic (Model: gatortron) 📊 Processing embeddings for data_85_quadratic (Model: roberta) 📊 Processing embeddings for data_85_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_85_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_85_quadratic (Model: e5_small) 📊 Processing embeddings for data_85_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_51_quadratic 📊 Processing embeddings for data_51_quadratic (Model: bert) 📊 Processing embeddings for data_51_quadratic (Model: ernie) 📊 Processing embeddings for data_51_quadratic (Model: gatortron) 📊 Processing embeddings for data_51_quadratic (Model: roberta) 📊 Processing embeddings for data_51_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_51_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_51_quadratic (Model: e5_small) 📊 Processing embeddings for data_51_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_90_quadratic 📊 Processing embeddings for data_90_quadratic (Model: bert) 📊 Processing embeddings for data_90_quadratic (Model: ernie) 📊 Processing embeddings for data_90_quadratic (Model: gatortron) 📊 Processing embeddings for data_90_quadratic (Model: roberta) 📊 Processing embeddings for data_90_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_90_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_90_quadratic (Model: e5_small) 📊 Processing embeddings for data_90_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_38_exponential 📊 Processing embeddings for data_38_exponential (Model: bert) 📊 Processing embeddings for data_38_exponential (Model: ernie) 📊 Processing embeddings for data_38_exponential (Model: gatortron) 📊 Processing embeddings for data_38_exponential (Model: roberta) 📊 Processing embeddings for data_38_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_38_exponential (Model: t5), flattening... 📊 Processing embeddings for data_38_exponential (Model: e5_small) 📊 Processing embeddings for data_38_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_81_exponential 📊 Processing embeddings for data_81_exponential (Model: bert) 📊 Processing embeddings for data_81_exponential (Model: ernie) 📊 Processing embeddings for data_81_exponential (Model: gatortron) 📊 Processing embeddings for data_81_exponential (Model: roberta) 📊 Processing embeddings for data_81_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_81_exponential (Model: t5), flattening... 📊 Processing embeddings for data_81_exponential (Model: e5_small) 📊 Processing embeddings for data_81_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_92_exponential 📊 Processing embeddings for data_92_exponential (Model: bert) 📊 Processing embeddings for data_92_exponential (Model: ernie) 📊 Processing embeddings for data_92_exponential (Model: gatortron) 📊 Processing embeddings for data_92_exponential (Model: roberta) 📊 Processing embeddings for data_92_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_92_exponential (Model: t5), flattening... 📊 Processing embeddings for data_92_exponential (Model: e5_small) 📊 Processing embeddings for data_92_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_62_quadratic 📊 Processing embeddings for data_62_quadratic (Model: bert) 📊 Processing embeddings for data_62_quadratic (Model: ernie) 📊 Processing embeddings for data_62_quadratic (Model: gatortron) 📊 Processing embeddings for data_62_quadratic (Model: roberta) 📊 Processing embeddings for data_62_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_62_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_62_quadratic (Model: e5_small) 📊 Processing embeddings for data_62_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_10_linear 📊 Processing embeddings for data_10_linear (Model: bert) 📊 Processing embeddings for data_10_linear (Model: ernie) 📊 Processing embeddings for data_10_linear (Model: gatortron) 📊 Processing embeddings for data_10_linear (Model: roberta) 📊 Processing embeddings for data_10_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_10_linear (Model: t5), flattening... 📊 Processing embeddings for data_10_linear (Model: e5_small) 📊 Processing embeddings for data_10_linear (Model: minilm) 🔍 Computing cluster similarity for: data_31_linear 📊 Processing embeddings for data_31_linear (Model: bert) 📊 Processing embeddings for data_31_linear (Model: ernie) 📊 Processing embeddings for data_31_linear (Model: gatortron) 📊 Processing embeddings for data_31_linear (Model: roberta) 📊 Processing embeddings for data_31_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_31_linear (Model: t5), flattening... 📊 Processing embeddings for data_31_linear (Model: e5_small) 📊 Processing embeddings for data_31_linear (Model: minilm) 🔍 Computing cluster similarity for: data_57_linear 📊 Processing embeddings for data_57_linear (Model: bert) 📊 Processing embeddings for data_57_linear (Model: ernie) 📊 Processing embeddings for data_57_linear (Model: gatortron) 📊 Processing embeddings for data_57_linear (Model: roberta) 📊 Processing embeddings for data_57_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_57_linear (Model: t5), flattening... 📊 Processing embeddings for data_57_linear (Model: e5_small) 📊 Processing embeddings for data_57_linear (Model: minilm) 🔍 Computing cluster similarity for: data_70_linear 📊 Processing embeddings for data_70_linear (Model: bert) 📊 Processing embeddings for data_70_linear (Model: ernie) 📊 Processing embeddings for data_70_linear (Model: gatortron) 📊 Processing embeddings for data_70_linear (Model: roberta) 📊 Processing embeddings for data_70_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_70_linear (Model: t5), flattening... 📊 Processing embeddings for data_70_linear (Model: e5_small) 📊 Processing embeddings for data_70_linear (Model: minilm) 🔍 Computing cluster similarity for: data_98_cubic 📊 Processing embeddings for data_98_cubic (Model: bert) 📊 Processing embeddings for data_98_cubic (Model: ernie) 📊 Processing embeddings for data_98_cubic (Model: gatortron) 📊 Processing embeddings for data_98_cubic (Model: roberta) 📊 Processing embeddings for data_98_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_98_cubic (Model: t5), flattening... 📊 Processing embeddings for data_98_cubic (Model: e5_small) 📊 Processing embeddings for data_98_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_23_cubic 📊 Processing embeddings for data_23_cubic (Model: bert) 📊 Processing embeddings for data_23_cubic (Model: ernie) 📊 Processing embeddings for data_23_cubic (Model: gatortron) 📊 Processing embeddings for data_23_cubic (Model: roberta) 📊 Processing embeddings for data_23_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_23_cubic (Model: t5), flattening... 📊 Processing embeddings for data_23_cubic (Model: e5_small) 📊 Processing embeddings for data_23_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_27_exponential 📊 Processing embeddings for data_27_exponential (Model: bert) 📊 Processing embeddings for data_27_exponential (Model: ernie) 📊 Processing embeddings for data_27_exponential (Model: gatortron) 📊 Processing embeddings for data_27_exponential (Model: roberta) 📊 Processing embeddings for data_27_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_27_exponential (Model: t5), flattening... 📊 Processing embeddings for data_27_exponential (Model: e5_small) 📊 Processing embeddings for data_27_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_61_exponential 📊 Processing embeddings for data_61_exponential (Model: bert) 📊 Processing embeddings for data_61_exponential (Model: ernie) 📊 Processing embeddings for data_61_exponential (Model: gatortron) 📊 Processing embeddings for data_61_exponential (Model: roberta) 📊 Processing embeddings for data_61_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_61_exponential (Model: t5), flattening... 📊 Processing embeddings for data_61_exponential (Model: e5_small) 📊 Processing embeddings for data_61_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_46_linear 📊 Processing embeddings for data_46_linear (Model: bert) 📊 Processing embeddings for data_46_linear (Model: ernie) 📊 Processing embeddings for data_46_linear (Model: gatortron) 📊 Processing embeddings for data_46_linear (Model: roberta) 📊 Processing embeddings for data_46_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_46_linear (Model: t5), flattening... 📊 Processing embeddings for data_46_linear (Model: e5_small) 📊 Processing embeddings for data_46_linear (Model: minilm) 🔍 Computing cluster similarity for: data_73_cubic 📊 Processing embeddings for data_73_cubic (Model: bert) 📊 Processing embeddings for data_73_cubic (Model: ernie) 📊 Processing embeddings for data_73_cubic (Model: gatortron) 📊 Processing embeddings for data_73_cubic (Model: roberta) 📊 Processing embeddings for data_73_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_73_cubic (Model: t5), flattening... 📊 Processing embeddings for data_73_cubic (Model: e5_small) 📊 Processing embeddings for data_73_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_53_quadratic 📊 Processing embeddings for data_53_quadratic (Model: bert) 📊 Processing embeddings for data_53_quadratic (Model: ernie) 📊 Processing embeddings for data_53_quadratic (Model: gatortron) 📊 Processing embeddings for data_53_quadratic (Model: roberta) 📊 Processing embeddings for data_53_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_53_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_53_quadratic (Model: e5_small) 📊 Processing embeddings for data_53_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_37_linear 📊 Processing embeddings for data_37_linear (Model: bert) 📊 Processing embeddings for data_37_linear (Model: ernie) 📊 Processing embeddings for data_37_linear (Model: gatortron) 📊 Processing embeddings for data_37_linear (Model: roberta) 📊 Processing embeddings for data_37_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_37_linear (Model: t5), flattening... 📊 Processing embeddings for data_37_linear (Model: e5_small) 📊 Processing embeddings for data_37_linear (Model: minilm) 🔍 Computing cluster similarity for: data_54_quadratic 📊 Processing embeddings for data_54_quadratic (Model: bert) 📊 Processing embeddings for data_54_quadratic (Model: ernie) 📊 Processing embeddings for data_54_quadratic (Model: gatortron) 📊 Processing embeddings for data_54_quadratic (Model: roberta) 📊 Processing embeddings for data_54_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_54_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_54_quadratic (Model: e5_small) 📊 Processing embeddings for data_54_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_20_quadratic 📊 Processing embeddings for data_20_quadratic (Model: bert) 📊 Processing embeddings for data_20_quadratic (Model: ernie) 📊 Processing embeddings for data_20_quadratic (Model: gatortron) 📊 Processing embeddings for data_20_quadratic (Model: roberta) 📊 Processing embeddings for data_20_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_20_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_20_quadratic (Model: e5_small) 📊 Processing embeddings for data_20_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_78_quadratic 📊 Processing embeddings for data_78_quadratic (Model: bert) 📊 Processing embeddings for data_78_quadratic (Model: ernie) 📊 Processing embeddings for data_78_quadratic (Model: gatortron) 📊 Processing embeddings for data_78_quadratic (Model: roberta) 📊 Processing embeddings for data_78_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_78_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_78_quadratic (Model: e5_small) 📊 Processing embeddings for data_78_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_82_quadratic 📊 Processing embeddings for data_82_quadratic (Model: bert) 📊 Processing embeddings for data_82_quadratic (Model: ernie) 📊 Processing embeddings for data_82_quadratic (Model: gatortron) 📊 Processing embeddings for data_82_quadratic (Model: roberta) 📊 Processing embeddings for data_82_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_82_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_82_quadratic (Model: e5_small) 📊 Processing embeddings for data_82_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_19_linear 📊 Processing embeddings for data_19_linear (Model: bert) 📊 Processing embeddings for data_19_linear (Model: ernie) 📊 Processing embeddings for data_19_linear (Model: gatortron) 📊 Processing embeddings for data_19_linear (Model: roberta) 📊 Processing embeddings for data_19_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_19_linear (Model: t5), flattening... 📊 Processing embeddings for data_19_linear (Model: e5_small) 📊 Processing embeddings for data_19_linear (Model: minilm) 🔍 Computing cluster similarity for: data_4_quadratic 📊 Processing embeddings for data_4_quadratic (Model: bert) 📊 Processing embeddings for data_4_quadratic (Model: ernie) 📊 Processing embeddings for data_4_quadratic (Model: gatortron) 📊 Processing embeddings for data_4_quadratic (Model: roberta) 📊 Processing embeddings for data_4_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_4_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_4_quadratic (Model: e5_small) 📊 Processing embeddings for data_4_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_8_quadratic 📊 Processing embeddings for data_8_quadratic (Model: bert) 📊 Processing embeddings for data_8_quadratic (Model: ernie) 📊 Processing embeddings for data_8_quadratic (Model: gatortron) 📊 Processing embeddings for data_8_quadratic (Model: roberta) 📊 Processing embeddings for data_8_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_8_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_8_quadratic (Model: e5_small) 📊 Processing embeddings for data_8_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_36_cubic 📊 Processing embeddings for data_36_cubic (Model: bert) 📊 Processing embeddings for data_36_cubic (Model: ernie) 📊 Processing embeddings for data_36_cubic (Model: gatortron) 📊 Processing embeddings for data_36_cubic (Model: roberta) 📊 Processing embeddings for data_36_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_36_cubic (Model: t5), flattening... 📊 Processing embeddings for data_36_cubic (Model: e5_small) 📊 Processing embeddings for data_36_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_24_exponential 📊 Processing embeddings for data_24_exponential (Model: bert) 📊 Processing embeddings for data_24_exponential (Model: ernie) 📊 Processing embeddings for data_24_exponential (Model: gatortron) 📊 Processing embeddings for data_24_exponential (Model: roberta) 📊 Processing embeddings for data_24_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_24_exponential (Model: t5), flattening... 📊 Processing embeddings for data_24_exponential (Model: e5_small) 📊 Processing embeddings for data_24_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_39_quadratic 📊 Processing embeddings for data_39_quadratic (Model: bert) 📊 Processing embeddings for data_39_quadratic (Model: ernie) 📊 Processing embeddings for data_39_quadratic (Model: gatortron) 📊 Processing embeddings for data_39_quadratic (Model: roberta) 📊 Processing embeddings for data_39_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_39_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_39_quadratic (Model: e5_small) 📊 Processing embeddings for data_39_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_60_cubic 📊 Processing embeddings for data_60_cubic (Model: bert) 📊 Processing embeddings for data_60_cubic (Model: ernie) 📊 Processing embeddings for data_60_cubic (Model: gatortron) 📊 Processing embeddings for data_60_cubic (Model: roberta) 📊 Processing embeddings for data_60_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_60_cubic (Model: t5), flattening... 📊 Processing embeddings for data_60_cubic (Model: e5_small) 📊 Processing embeddings for data_60_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_1_quadratic 📊 Processing embeddings for data_1_quadratic (Model: bert) 📊 Processing embeddings for data_1_quadratic (Model: ernie) 📊 Processing embeddings for data_1_quadratic (Model: gatortron) 📊 Processing embeddings for data_1_quadratic (Model: roberta) 📊 Processing embeddings for data_1_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_1_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_1_quadratic (Model: e5_small) 📊 Processing embeddings for data_1_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_41_exponential 📊 Processing embeddings for data_41_exponential (Model: bert) 📊 Processing embeddings for data_41_exponential (Model: ernie) 📊 Processing embeddings for data_41_exponential (Model: gatortron) 📊 Processing embeddings for data_41_exponential (Model: roberta) 📊 Processing embeddings for data_41_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_41_exponential (Model: t5), flattening... 📊 Processing embeddings for data_41_exponential (Model: e5_small) 📊 Processing embeddings for data_41_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_32_linear 📊 Processing embeddings for data_32_linear (Model: bert) 📊 Processing embeddings for data_32_linear (Model: ernie) 📊 Processing embeddings for data_32_linear (Model: gatortron) 📊 Processing embeddings for data_32_linear (Model: roberta) 📊 Processing embeddings for data_32_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_32_linear (Model: t5), flattening... 📊 Processing embeddings for data_32_linear (Model: e5_small) 📊 Processing embeddings for data_32_linear (Model: minilm) 🔍 Computing cluster similarity for: data_66_linear 📊 Processing embeddings for data_66_linear (Model: bert) 📊 Processing embeddings for data_66_linear (Model: ernie) 📊 Processing embeddings for data_66_linear (Model: gatortron) 📊 Processing embeddings for data_66_linear (Model: roberta) 📊 Processing embeddings for data_66_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_66_linear (Model: t5), flattening... 📊 Processing embeddings for data_66_linear (Model: e5_small) 📊 Processing embeddings for data_66_linear (Model: minilm) 🔍 Computing cluster similarity for: data_49_linear 📊 Processing embeddings for data_49_linear (Model: bert) 📊 Processing embeddings for data_49_linear (Model: ernie) 📊 Processing embeddings for data_49_linear (Model: gatortron) 📊 Processing embeddings for data_49_linear (Model: roberta) 📊 Processing embeddings for data_49_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_49_linear (Model: t5), flattening... 📊 Processing embeddings for data_49_linear (Model: e5_small) 🔍 Computing cluster similarity for: data_56_linear 📊 Processing embeddings for data_56_linear (Model: bert) 📊 Processing embeddings for data_56_linear (Model: ernie) 📊 Processing embeddings for data_56_linear (Model: gatortron) 📊 Processing embeddings for data_56_linear (Model: roberta) 📊 Processing embeddings for data_56_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_56_linear (Model: t5), flattening... 📊 Processing embeddings for data_56_linear (Model: e5_small) 📊 Processing embeddings for data_56_linear (Model: minilm) 🔍 Computing cluster similarity for: data_63_exponential 📊 Processing embeddings for data_63_exponential (Model: bert) 📊 Processing embeddings for data_63_exponential (Model: ernie) 📊 Processing embeddings for data_63_exponential (Model: gatortron) 📊 Processing embeddings for data_63_exponential (Model: roberta) 📊 Processing embeddings for data_63_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_63_exponential (Model: t5), flattening... 📊 Processing embeddings for data_63_exponential (Model: e5_small) 📊 Processing embeddings for data_63_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_80_cubic 📊 Processing embeddings for data_80_cubic (Model: bert) 📊 Processing embeddings for data_80_cubic (Model: ernie) 📊 Processing embeddings for data_80_cubic (Model: gatortron) 📊 Processing embeddings for data_80_cubic (Model: roberta) 📊 Processing embeddings for data_80_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_80_cubic (Model: t5), flattening... 📊 Processing embeddings for data_80_cubic (Model: e5_small) 🔍 Computing cluster similarity for: data_12_exponential 📊 Processing embeddings for data_12_exponential (Model: bert) 📊 Processing embeddings for data_12_exponential (Model: ernie) 📊 Processing embeddings for data_12_exponential (Model: gatortron) 📊 Processing embeddings for data_12_exponential (Model: roberta) 📊 Processing embeddings for data_12_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_12_exponential (Model: t5), flattening... 📊 Processing embeddings for data_12_exponential (Model: e5_small) 📊 Processing embeddings for data_12_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_3_quadratic 📊 Processing embeddings for data_3_quadratic (Model: bert) 📊 Processing embeddings for data_3_quadratic (Model: ernie) 📊 Processing embeddings for data_3_quadratic (Model: gatortron) 📊 Processing embeddings for data_3_quadratic (Model: roberta) 📊 Processing embeddings for data_3_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_3_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_3_quadratic (Model: e5_small) 📊 Processing embeddings for data_3_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_64_exponential 📊 Processing embeddings for data_64_exponential (Model: bert) 📊 Processing embeddings for data_64_exponential (Model: ernie) 📊 Processing embeddings for data_64_exponential (Model: gatortron) 📊 Processing embeddings for data_64_exponential (Model: roberta) 📊 Processing embeddings for data_64_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_64_exponential (Model: t5), flattening... 📊 Processing embeddings for data_64_exponential (Model: e5_small) 📊 Processing embeddings for data_64_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_35_cubic 📊 Processing embeddings for data_35_cubic (Model: bert) 📊 Processing embeddings for data_35_cubic (Model: ernie) 📊 Processing embeddings for data_35_cubic (Model: gatortron) 📊 Processing embeddings for data_35_cubic (Model: roberta) 📊 Processing embeddings for data_35_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_35_cubic (Model: t5), flattening... 📊 Processing embeddings for data_35_cubic (Model: e5_small) 📊 Processing embeddings for data_35_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_6_linear 📊 Processing embeddings for data_6_linear (Model: bert) 📊 Processing embeddings for data_6_linear (Model: ernie) 📊 Processing embeddings for data_6_linear (Model: gatortron) 📊 Processing embeddings for data_6_linear (Model: roberta) 📊 Processing embeddings for data_6_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_6_linear (Model: t5), flattening... 📊 Processing embeddings for data_6_linear (Model: e5_small) 📊 Processing embeddings for data_6_linear (Model: minilm) 🔍 Computing cluster similarity for: data_93_exponential 📊 Processing embeddings for data_93_exponential (Model: bert) 📊 Processing embeddings for data_93_exponential (Model: ernie) 📊 Processing embeddings for data_93_exponential (Model: gatortron) 📊 Processing embeddings for data_93_exponential (Model: roberta) 📊 Processing embeddings for data_93_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_93_exponential (Model: t5), flattening... 📊 Processing embeddings for data_93_exponential (Model: e5_small) 📊 Processing embeddings for data_93_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_29_linear 📊 Processing embeddings for data_29_linear (Model: bert) 📊 Processing embeddings for data_29_linear (Model: ernie) 📊 Processing embeddings for data_29_linear (Model: gatortron) 📊 Processing embeddings for data_29_linear (Model: roberta) 📊 Processing embeddings for data_29_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_29_linear (Model: t5), flattening... 📊 Processing embeddings for data_29_linear (Model: e5_small) 📊 Processing embeddings for data_29_linear (Model: minilm) 🔍 Computing cluster similarity for: data_76_cubic 📊 Processing embeddings for data_76_cubic (Model: bert) 📊 Processing embeddings for data_76_cubic (Model: ernie) 📊 Processing embeddings for data_76_cubic (Model: gatortron) 📊 Processing embeddings for data_76_cubic (Model: roberta) 📊 Processing embeddings for data_76_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_76_cubic (Model: t5), flattening... 📊 Processing embeddings for data_76_cubic (Model: e5_small) 📊 Processing embeddings for data_76_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_50_exponential 📊 Processing embeddings for data_50_exponential (Model: bert) 📊 Processing embeddings for data_50_exponential (Model: ernie) 📊 Processing embeddings for data_50_exponential (Model: gatortron) 📊 Processing embeddings for data_50_exponential (Model: roberta) 📊 Processing embeddings for data_50_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_50_exponential (Model: t5), flattening... 📊 Processing embeddings for data_50_exponential (Model: e5_small) 📊 Processing embeddings for data_50_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_9_exponential 📊 Processing embeddings for data_9_exponential (Model: bert) 📊 Processing embeddings for data_9_exponential (Model: ernie) 📊 Processing embeddings for data_9_exponential (Model: gatortron) 📊 Processing embeddings for data_9_exponential (Model: roberta) 📊 Processing embeddings for data_9_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_9_exponential (Model: t5), flattening... 📊 Processing embeddings for data_9_exponential (Model: e5_small) 📊 Processing embeddings for data_9_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_59_quadratic 📊 Processing embeddings for data_59_quadratic (Model: bert) 📊 Processing embeddings for data_59_quadratic (Model: ernie) 📊 Processing embeddings for data_59_quadratic (Model: gatortron) 📊 Processing embeddings for data_59_quadratic (Model: roberta) 📊 Processing embeddings for data_59_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_59_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_59_quadratic (Model: e5_small) 📊 Processing embeddings for data_59_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_97_quadratic 📊 Processing embeddings for data_97_quadratic (Model: bert) 📊 Processing embeddings for data_97_quadratic (Model: ernie) 📊 Processing embeddings for data_97_quadratic (Model: gatortron) 📊 Processing embeddings for data_97_quadratic (Model: roberta) 📊 Processing embeddings for data_97_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_97_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_97_quadratic (Model: e5_small) 📊 Processing embeddings for data_97_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_45_cubic 📊 Processing embeddings for data_45_cubic (Model: bert) 📊 Processing embeddings for data_45_cubic (Model: ernie) 📊 Processing embeddings for data_45_cubic (Model: gatortron) 📊 Processing embeddings for data_45_cubic (Model: roberta) 📊 Processing embeddings for data_45_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_45_cubic (Model: t5), flattening... 📊 Processing embeddings for data_45_cubic (Model: e5_small) 📊 Processing embeddings for data_45_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_18_quadratic 📊 Processing embeddings for data_18_quadratic (Model: bert) 📊 Processing embeddings for data_18_quadratic (Model: ernie) 📊 Processing embeddings for data_18_quadratic (Model: gatortron) 📊 Processing embeddings for data_18_quadratic (Model: roberta) 📊 Processing embeddings for data_18_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_18_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_18_quadratic (Model: e5_small) 📊 Processing embeddings for data_18_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_69_linear 📊 Processing embeddings for data_69_linear (Model: bert) 📊 Processing embeddings for data_69_linear (Model: ernie) 📊 Processing embeddings for data_69_linear (Model: gatortron) 📊 Processing embeddings for data_69_linear (Model: roberta) 📊 Processing embeddings for data_69_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_69_linear (Model: t5), flattening... 📊 Processing embeddings for data_69_linear (Model: e5_small) 📊 Processing embeddings for data_69_linear (Model: minilm) 🔍 Computing cluster similarity for: data_67_linear 📊 Processing embeddings for data_67_linear (Model: bert) 📊 Processing embeddings for data_67_linear (Model: ernie) 📊 Processing embeddings for data_67_linear (Model: gatortron) 📊 Processing embeddings for data_67_linear (Model: roberta) 📊 Processing embeddings for data_67_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_67_linear (Model: t5), flattening... 📊 Processing embeddings for data_67_linear (Model: e5_small) 📊 Processing embeddings for data_67_linear (Model: minilm) 🔍 Computing cluster similarity for: data_58_quadratic 📊 Processing embeddings for data_58_quadratic (Model: bert) 📊 Processing embeddings for data_58_quadratic (Model: ernie) 📊 Processing embeddings for data_58_quadratic (Model: gatortron) 📊 Processing embeddings for data_58_quadratic (Model: roberta) 📊 Processing embeddings for data_58_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_58_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_58_quadratic (Model: e5_small) 📊 Processing embeddings for data_58_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_95_quadratic 📊 Processing embeddings for data_95_quadratic (Model: bert) 📊 Processing embeddings for data_95_quadratic (Model: ernie) 📊 Processing embeddings for data_95_quadratic (Model: gatortron) 📊 Processing embeddings for data_95_quadratic (Model: roberta) 📊 Processing embeddings for data_95_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_95_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_95_quadratic (Model: e5_small) 📊 Processing embeddings for data_95_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_11_exponential 📊 Processing embeddings for data_11_exponential (Model: bert) 📊 Processing embeddings for data_11_exponential (Model: ernie) 📊 Processing embeddings for data_11_exponential (Model: gatortron) 📊 Processing embeddings for data_11_exponential (Model: roberta) 📊 Processing embeddings for data_11_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_11_exponential (Model: t5), flattening... 📊 Processing embeddings for data_11_exponential (Model: e5_small) 📊 Processing embeddings for data_11_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_87_quadratic 📊 Processing embeddings for data_87_quadratic (Model: bert) 📊 Processing embeddings for data_87_quadratic (Model: ernie) 📊 Processing embeddings for data_87_quadratic (Model: gatortron) 📊 Processing embeddings for data_87_quadratic (Model: roberta) 📊 Processing embeddings for data_87_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_87_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_87_quadratic (Model: e5_small) 📊 Processing embeddings for data_87_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_2_linear 📊 Processing embeddings for data_2_linear (Model: bert) 📊 Processing embeddings for data_2_linear (Model: ernie) 📊 Processing embeddings for data_2_linear (Model: gatortron) 📊 Processing embeddings for data_2_linear (Model: roberta) 📊 Processing embeddings for data_2_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_2_linear (Model: t5), flattening... 📊 Processing embeddings for data_2_linear (Model: e5_small) 📊 Processing embeddings for data_2_linear (Model: minilm) 🔍 Computing cluster similarity for: data_47_cubic 📊 Processing embeddings for data_47_cubic (Model: bert) 📊 Processing embeddings for data_47_cubic (Model: ernie) 📊 Processing embeddings for data_47_cubic (Model: gatortron) 📊 Processing embeddings for data_47_cubic (Model: roberta) 📊 Processing embeddings for data_47_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_47_cubic (Model: t5), flattening... 📊 Processing embeddings for data_47_cubic (Model: e5_small) 📊 Processing embeddings for data_47_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_75_exponential 📊 Processing embeddings for data_75_exponential (Model: bert) 📊 Processing embeddings for data_75_exponential (Model: ernie) 📊 Processing embeddings for data_75_exponential (Model: gatortron) 📊 Processing embeddings for data_75_exponential (Model: roberta) 📊 Processing embeddings for data_75_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_75_exponential (Model: t5), flattening... 📊 Processing embeddings for data_75_exponential (Model: e5_small) 📊 Processing embeddings for data_75_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_83_exponential 📊 Processing embeddings for data_83_exponential (Model: bert) 📊 Processing embeddings for data_83_exponential (Model: ernie) 📊 Processing embeddings for data_83_exponential (Model: gatortron) 📊 Processing embeddings for data_83_exponential (Model: roberta) 📊 Processing embeddings for data_83_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_83_exponential (Model: t5), flattening... 📊 Processing embeddings for data_83_exponential (Model: e5_small) 📊 Processing embeddings for data_83_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_99_exponential 📊 Processing embeddings for data_99_exponential (Model: bert) 📊 Processing embeddings for data_99_exponential (Model: ernie) 📊 Processing embeddings for data_99_exponential (Model: gatortron) 📊 Processing embeddings for data_99_exponential (Model: roberta) 📊 Processing embeddings for data_99_exponential (Model: t5) ⚠️ Warning: Found 3D embeddings for data_99_exponential (Model: t5), flattening... 📊 Processing embeddings for data_99_exponential (Model: e5_small) 📊 Processing embeddings for data_99_exponential (Model: minilm) 🔍 Computing cluster similarity for: data_7_cubic 📊 Processing embeddings for data_7_cubic (Model: bert) 📊 Processing embeddings for data_7_cubic (Model: ernie) 📊 Processing embeddings for data_7_cubic (Model: gatortron) 📊 Processing embeddings for data_7_cubic (Model: roberta) 📊 Processing embeddings for data_7_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_7_cubic (Model: t5), flattening... 📊 Processing embeddings for data_7_cubic (Model: e5_small) 📊 Processing embeddings for data_7_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_15_quadratic 📊 Processing embeddings for data_15_quadratic (Model: bert) 📊 Processing embeddings for data_15_quadratic (Model: ernie) 📊 Processing embeddings for data_15_quadratic (Model: gatortron) 📊 Processing embeddings for data_15_quadratic (Model: roberta) 📊 Processing embeddings for data_15_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_15_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_15_quadratic (Model: e5_small) 📊 Processing embeddings for data_15_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_71_cubic 📊 Processing embeddings for data_71_cubic (Model: bert) 📊 Processing embeddings for data_71_cubic (Model: ernie) 📊 Processing embeddings for data_71_cubic (Model: gatortron) 📊 Processing embeddings for data_71_cubic (Model: roberta) 📊 Processing embeddings for data_71_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_71_cubic (Model: t5), flattening... 📊 Processing embeddings for data_71_cubic (Model: e5_small) 📊 Processing embeddings for data_71_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_77_cubic 📊 Processing embeddings for data_77_cubic (Model: bert) 📊 Processing embeddings for data_77_cubic (Model: ernie) 📊 Processing embeddings for data_77_cubic (Model: gatortron) 📊 Processing embeddings for data_77_cubic (Model: roberta) 📊 Processing embeddings for data_77_cubic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_77_cubic (Model: t5), flattening... 📊 Processing embeddings for data_77_cubic (Model: e5_small) 📊 Processing embeddings for data_77_cubic (Model: minilm) 🔍 Computing cluster similarity for: data_86_linear 📊 Processing embeddings for data_86_linear (Model: bert) 📊 Processing embeddings for data_86_linear (Model: ernie) 📊 Processing embeddings for data_86_linear (Model: gatortron) 📊 Processing embeddings for data_86_linear (Model: roberta) 📊 Processing embeddings for data_86_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_86_linear (Model: t5), flattening... 📊 Processing embeddings for data_86_linear (Model: e5_small) 📊 Processing embeddings for data_86_linear (Model: minilm) 🔍 Computing cluster similarity for: data_26_quadratic 📊 Processing embeddings for data_26_quadratic (Model: bert) 📊 Processing embeddings for data_26_quadratic (Model: ernie) 📊 Processing embeddings for data_26_quadratic (Model: gatortron) 📊 Processing embeddings for data_26_quadratic (Model: roberta) 📊 Processing embeddings for data_26_quadratic (Model: t5) ⚠️ Warning: Found 3D embeddings for data_26_quadratic (Model: t5), flattening... 📊 Processing embeddings for data_26_quadratic (Model: e5_small) 📊 Processing embeddings for data_26_quadratic (Model: minilm) 🔍 Computing cluster similarity for: data_28_linear 📊 Processing embeddings for data_28_linear (Model: bert) 📊 Processing embeddings for data_28_linear (Model: ernie) 📊 Processing embeddings for data_28_linear (Model: gatortron) 📊 Processing embeddings for data_28_linear (Model: roberta) 📊 Processing embeddings for data_28_linear (Model: t5) ⚠️ Warning: Found 3D embeddings for data_28_linear (Model: t5), flattening... 📊 Processing embeddings for data_28_linear (Model: e5_small) 📊 Processing embeddings for data_28_linear (Model: minilm) 📂 Processing /home/abbasali/synthetic data/generated_data_llama -> /home/abbasali/synthetic data/embeddings_llama ✅ Loaded 79 CSV datasets from /home/abbasali/synthetic data/generated_data_llama. ✅ Loaded embeddings for data_79_quadratic (Model: llama) ✅ Loaded embeddings for data_91_linear (Model: llama) ✅ Loaded embeddings for data_44_exponential (Model: llama) ✅ Loaded embeddings for data_42_linear (Model: llama) ✅ Loaded embeddings for data_43_linear (Model: llama) ✅ Loaded embeddings for data_48_cubic (Model: llama) ✅ Loaded embeddings for data_13_cubic (Model: llama) ✅ Loaded embeddings for data_33_quadratic (Model: llama) ✅ Loaded embeddings for data_84_cubic (Model: llama) ✅ Loaded embeddings for data_74_linear (Model: llama) ✅ Loaded embeddings for data_40_exponential (Model: llama) ✅ Loaded embeddings for data_17_exponential (Model: llama) ✅ Loaded embeddings for data_30_cubic (Model: llama) ✅ Loaded embeddings for data_34_cubic (Model: llama) ✅ Loaded embeddings for data_52_cubic (Model: llama) ✅ Loaded embeddings for data_72_linear (Model: llama) ✅ Loaded embeddings for data_89_linear (Model: llama) ✅ Loaded embeddings for data_55_linear (Model: llama) ✅ Loaded embeddings for data_22_quadratic (Model: llama) ✅ Loaded embeddings for data_65_exponential (Model: llama) ✅ Loaded embeddings for data_14_linear (Model: llama) ✅ Loaded embeddings for data_25_exponential (Model: llama) ✅ Loaded embeddings for data_94_quadratic (Model: llama) ✅ Loaded embeddings for data_96_cubic (Model: llama) ✅ Loaded embeddings for data_85_quadratic (Model: llama) ✅ Loaded embeddings for data_51_quadratic (Model: llama) ✅ Loaded embeddings for data_90_quadratic (Model: llama) ✅ Loaded embeddings for data_38_exponential (Model: llama) ✅ Loaded embeddings for data_81_exponential (Model: llama) ✅ Loaded embeddings for data_62_quadratic (Model: llama) ✅ Loaded embeddings for data_10_linear (Model: llama) ✅ Loaded embeddings for data_31_linear (Model: llama) ✅ Loaded embeddings for data_70_linear (Model: llama) ✅ Loaded embeddings for data_98_cubic (Model: llama) ✅ Loaded embeddings for data_27_exponential (Model: llama) ✅ Loaded embeddings for data_53_quadratic (Model: llama) ✅ Loaded embeddings for data_37_linear (Model: llama) ✅ Loaded embeddings for data_54_quadratic (Model: llama) ✅ Loaded embeddings for data_20_quadratic (Model: llama) ✅ Loaded embeddings for data_78_quadratic (Model: llama) ✅ Loaded embeddings for data_19_linear (Model: llama) ✅ Loaded embeddings for data_4_quadratic (Model: llama) ✅ Loaded embeddings for data_36_cubic (Model: llama) ✅ Loaded embeddings for data_24_exponential (Model: llama) ✅ Loaded embeddings for data_39_quadratic (Model: llama) ✅ Loaded embeddings for data_1_quadratic (Model: llama) ✅ Loaded embeddings for data_41_exponential (Model: llama) ✅ Loaded embeddings for data_32_linear (Model: llama) ✅ Loaded embeddings for data_66_linear (Model: llama) ✅ Loaded embeddings for data_49_linear (Model: llama) ✅ Loaded embeddings for data_56_linear (Model: llama) ✅ Loaded embeddings for data_63_exponential (Model: llama) ✅ Loaded embeddings for data_80_cubic (Model: llama) ✅ Loaded embeddings for data_12_exponential (Model: llama) ✅ Loaded embeddings for data_3_quadratic (Model: llama) ✅ Loaded embeddings for data_64_exponential (Model: llama) ✅ Loaded embeddings for data_35_cubic (Model: llama) ✅ Loaded embeddings for data_6_linear (Model: llama) ✅ Loaded embeddings for data_93_exponential (Model: llama) ✅ Loaded embeddings for data_76_cubic (Model: llama) ✅ Loaded embeddings for data_50_exponential (Model: llama) ✅ Loaded embeddings for data_9_exponential (Model: llama) ✅ Loaded embeddings for data_59_quadratic (Model: llama) ✅ Loaded embeddings for data_97_quadratic (Model: llama) ✅ Loaded embeddings for data_45_cubic (Model: llama) ✅ Loaded embeddings for data_18_quadratic (Model: llama) ✅ Loaded embeddings for data_69_linear (Model: llama) ✅ Loaded embeddings for data_58_quadratic (Model: llama) ✅ Loaded embeddings for data_11_exponential (Model: llama) ✅ Loaded embeddings for data_87_quadratic (Model: llama) ✅ Loaded embeddings for data_47_cubic (Model: llama) ✅ Loaded embeddings for data_75_exponential (Model: llama) ✅ Loaded embeddings for data_83_exponential (Model: llama) ✅ Loaded embeddings for data_99_exponential (Model: llama) ✅ Loaded embeddings for data_7_cubic (Model: llama) ✅ Loaded embeddings for data_15_quadratic (Model: llama) ✅ Loaded embeddings for data_71_cubic (Model: llama) ✅ Loaded embeddings for data_77_cubic (Model: llama) ⚠️ WARNING: Missing embeddings for data_26_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_79_quadratic 📊 Processing embeddings for data_79_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_91_linear 📊 Processing embeddings for data_91_linear (Model: llama) 🔍 Computing cluster similarity for: data_44_exponential 📊 Processing embeddings for data_44_exponential (Model: llama) 🔍 Computing cluster similarity for: data_42_linear 📊 Processing embeddings for data_42_linear (Model: llama) 🔍 Computing cluster similarity for: data_43_linear 📊 Processing embeddings for data_43_linear (Model: llama) 🔍 Computing cluster similarity for: data_48_cubic 📊 Processing embeddings for data_48_cubic (Model: llama) 🔍 Computing cluster similarity for: data_13_cubic 📊 Processing embeddings for data_13_cubic (Model: llama) 🔍 Computing cluster similarity for: data_33_quadratic 📊 Processing embeddings for data_33_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_84_cubic 📊 Processing embeddings for data_84_cubic (Model: llama) 🔍 Computing cluster similarity for: data_74_linear 📊 Processing embeddings for data_74_linear (Model: llama) 🔍 Computing cluster similarity for: data_40_exponential 📊 Processing embeddings for data_40_exponential (Model: llama) 🔍 Computing cluster similarity for: data_17_exponential 📊 Processing embeddings for data_17_exponential (Model: llama) 🔍 Computing cluster similarity for: data_30_cubic 📊 Processing embeddings for data_30_cubic (Model: llama) 🔍 Computing cluster similarity for: data_34_cubic 📊 Processing embeddings for data_34_cubic (Model: llama) 🔍 Computing cluster similarity for: data_52_cubic 📊 Processing embeddings for data_52_cubic (Model: llama) 🔍 Computing cluster similarity for: data_72_linear 📊 Processing embeddings for data_72_linear (Model: llama) 🔍 Computing cluster similarity for: data_89_linear 📊 Processing embeddings for data_89_linear (Model: llama) 🔍 Computing cluster similarity for: data_55_linear 📊 Processing embeddings for data_55_linear (Model: llama) 🔍 Computing cluster similarity for: data_22_quadratic 📊 Processing embeddings for data_22_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_65_exponential 📊 Processing embeddings for data_65_exponential (Model: llama) 🔍 Computing cluster similarity for: data_14_linear 📊 Processing embeddings for data_14_linear (Model: llama) 🔍 Computing cluster similarity for: data_25_exponential 📊 Processing embeddings for data_25_exponential (Model: llama) 🔍 Computing cluster similarity for: data_94_quadratic 📊 Processing embeddings for data_94_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_96_cubic 📊 Processing embeddings for data_96_cubic (Model: llama) 🔍 Computing cluster similarity for: data_85_quadratic 📊 Processing embeddings for data_85_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_51_quadratic 📊 Processing embeddings for data_51_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_90_quadratic 📊 Processing embeddings for data_90_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_38_exponential 📊 Processing embeddings for data_38_exponential (Model: llama) 🔍 Computing cluster similarity for: data_81_exponential 📊 Processing embeddings for data_81_exponential (Model: llama) 🔍 Computing cluster similarity for: data_62_quadratic 📊 Processing embeddings for data_62_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_10_linear 📊 Processing embeddings for data_10_linear (Model: llama) 🔍 Computing cluster similarity for: data_31_linear 📊 Processing embeddings for data_31_linear (Model: llama) 🔍 Computing cluster similarity for: data_70_linear 📊 Processing embeddings for data_70_linear (Model: llama) 🔍 Computing cluster similarity for: data_98_cubic 📊 Processing embeddings for data_98_cubic (Model: llama) 🔍 Computing cluster similarity for: data_27_exponential 📊 Processing embeddings for data_27_exponential (Model: llama) 🔍 Computing cluster similarity for: data_53_quadratic 📊 Processing embeddings for data_53_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_37_linear 📊 Processing embeddings for data_37_linear (Model: llama) 🔍 Computing cluster similarity for: data_54_quadratic 📊 Processing embeddings for data_54_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_20_quadratic 📊 Processing embeddings for data_20_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_78_quadratic 📊 Processing embeddings for data_78_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_19_linear 📊 Processing embeddings for data_19_linear (Model: llama) 🔍 Computing cluster similarity for: data_4_quadratic 📊 Processing embeddings for data_4_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_36_cubic 📊 Processing embeddings for data_36_cubic (Model: llama) 🔍 Computing cluster similarity for: data_24_exponential 📊 Processing embeddings for data_24_exponential (Model: llama) 🔍 Computing cluster similarity for: data_39_quadratic 📊 Processing embeddings for data_39_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_1_quadratic 📊 Processing embeddings for data_1_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_41_exponential 📊 Processing embeddings for data_41_exponential (Model: llama) 🔍 Computing cluster similarity for: data_32_linear 📊 Processing embeddings for data_32_linear (Model: llama) 🔍 Computing cluster similarity for: data_66_linear 📊 Processing embeddings for data_66_linear (Model: llama) 🔍 Computing cluster similarity for: data_49_linear 📊 Processing embeddings for data_49_linear (Model: llama) 🔍 Computing cluster similarity for: data_56_linear 📊 Processing embeddings for data_56_linear (Model: llama) 🔍 Computing cluster similarity for: data_63_exponential 📊 Processing embeddings for data_63_exponential (Model: llama) 🔍 Computing cluster similarity for: data_80_cubic 📊 Processing embeddings for data_80_cubic (Model: llama) 🔍 Computing cluster similarity for: data_12_exponential 📊 Processing embeddings for data_12_exponential (Model: llama) 🔍 Computing cluster similarity for: data_3_quadratic 📊 Processing embeddings for data_3_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_64_exponential 📊 Processing embeddings for data_64_exponential (Model: llama) 🔍 Computing cluster similarity for: data_35_cubic 📊 Processing embeddings for data_35_cubic (Model: llama) 🔍 Computing cluster similarity for: data_6_linear 📊 Processing embeddings for data_6_linear (Model: llama) 🔍 Computing cluster similarity for: data_93_exponential 📊 Processing embeddings for data_93_exponential (Model: llama) 🔍 Computing cluster similarity for: data_76_cubic 📊 Processing embeddings for data_76_cubic (Model: llama) 🔍 Computing cluster similarity for: data_50_exponential 📊 Processing embeddings for data_50_exponential (Model: llama) 🔍 Computing cluster similarity for: data_9_exponential 📊 Processing embeddings for data_9_exponential (Model: llama) 🔍 Computing cluster similarity for: data_59_quadratic 📊 Processing embeddings for data_59_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_97_quadratic 📊 Processing embeddings for data_97_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_45_cubic 📊 Processing embeddings for data_45_cubic (Model: llama) 🔍 Computing cluster similarity for: data_18_quadratic 📊 Processing embeddings for data_18_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_69_linear 📊 Processing embeddings for data_69_linear (Model: llama) 🔍 Computing cluster similarity for: data_58_quadratic 📊 Processing embeddings for data_58_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_11_exponential 📊 Processing embeddings for data_11_exponential (Model: llama) 🔍 Computing cluster similarity for: data_87_quadratic 📊 Processing embeddings for data_87_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_47_cubic 📊 Processing embeddings for data_47_cubic (Model: llama) 🔍 Computing cluster similarity for: data_75_exponential 📊 Processing embeddings for data_75_exponential (Model: llama) 🔍 Computing cluster similarity for: data_83_exponential 📊 Processing embeddings for data_83_exponential (Model: llama) 🔍 Computing cluster similarity for: data_99_exponential 📊 Processing embeddings for data_99_exponential (Model: llama) 🔍 Computing cluster similarity for: data_7_cubic 📊 Processing embeddings for data_7_cubic (Model: llama) 🔍 Computing cluster similarity for: data_15_quadratic 📊 Processing embeddings for data_15_quadratic (Model: llama) 🔍 Computing cluster similarity for: data_71_cubic 📊 Processing embeddings for data_71_cubic (Model: llama) 🔍 Computing cluster similarity for: data_77_cubic 📊 Processing embeddings for data_77_cubic (Model: llama) 🔍 Computing cluster similarity for: data_26_quadratic ✅ Results saved.
In [38]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from scipy.stats import sem
# ✅ Load the results CSV file
RESULTS_FILE = "cluster_similarity_results.csv"
df = pd.read_csv(RESULTS_FILE)
# ✅ Ensure necessary columns exist
required_columns = {"Model", "Function Type", "Distance Correlation", "Cosine Similarity"}
if not required_columns.issubset(df.columns):
raise ValueError(f"❌ ERROR: Missing required columns in {RESULTS_FILE}!")
# ✅ Compute mean and SEM for Distance Correlation and Cosine Similarity, grouped by Function Type and Model
stats_df = df.groupby(["Function Type", "Model"])[["Distance Correlation", "Cosine Similarity"]].agg(["mean", sem]).reset_index()
# ✅ Rename columns for easier access
stats_df.columns = ["Function Type", "Model", "Distance Mean", "Distance SEM", "Cosine Mean", "Cosine SEM"]
# ✅ Set visualization style
sns.set(style="whitegrid")
# ✅ Create plots for each function type
function_types = stats_df["Function Type"].unique()
for func_type in function_types:
for metric, metric_name, y_min in zip(
["Distance Mean", "Cosine Mean"],
["Distance Correlation", "Cosine Similarity"],
[None, 0.62] # Distance y-axis auto, Cosine starts at 0.5
):
subset = stats_df[stats_df["Function Type"] == func_type]
# ✅ Sort by mean in descending order
subset = subset.sort_values(by=metric, ascending=False)
# Select the corresponding SEM
sem_column = "Distance SEM" if metric == "Distance Mean" else "Cosine SEM"
# ✅ Dynamically adjust figure size based on number of models
fig_height = max(6, len(subset) * 0.8) # Ensures full graph without scrolling
plt.figure(figsize=(12, fig_height))
# Scatter plot for mean values
plt.scatter(subset["Model"], subset[metric], color="red", label="Mean", zorder=3)
# Error bars for SEM
plt.errorbar(subset["Model"], subset[metric], yerr=subset[sem_column], fmt="none", capsize=5, color="black", zorder=2)
# ✅ Add text annotations for Mean ± SEM **below the x-axis labels**
for i, row in subset.iterrows():
plt.text(
row["Model"],
row[metric] - row[sem_column] - 0.02, # Offset text slightly below error bar
f"{row[metric]:.2f} ± {row[sem_column]:.2f}",
ha="center",
fontsize=12,
fontweight="bold",
color="black"
)
# ✅ Set labels and title with font size 18
plt.xlabel("Model", fontsize=18)
plt.ylabel(f"{metric_name} Mean", fontsize=18)
plt.title(f"{metric_name} Across Models ({func_type.capitalize()})", fontsize=18)
# ✅ Set y-axis limits (Cosine starts at 0.5, Distance auto)
if y_min is not None:
plt.ylim(y_min, None)
# ✅ Set font sizes for ticks
plt.xticks(fontsize=18, rotation=45)
plt.yticks(fontsize=18)
# ✅ Add legend
plt.legend(fontsize=18, loc="upper right")
plt.grid(True, linestyle="--", alpha=0.6)
plt.tight_layout()
# ✅ Insert blank space between plots
plt.subplots_adjust(bottom=0.2)
plt.show()
In [39]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import scipy.stats as stats
# ✅ Load the results file
RESULTS_FILE = "cluster_similarity_results.csv"
results_df = pd.read_csv(RESULTS_FILE)
# ✅ Ensure that the dataset column is properly formatted
if "Dataset" in results_df.columns:
results_df["Function Type"] = results_df["Dataset"].str.extract(r'_(linear|exponential|quadratic|cubic)')
else:
raise ValueError("❌ ERROR: 'Dataset' column missing from results_df!")
# ✅ Function to calculate mean, SEM, and 95% confidence intervals
def compute_summary(df, metric):
"""Compute mean, standard error, and 95% confidence interval for a given metric."""
summary_df = df.groupby(["Model", "Function Type"])[metric].agg(["mean", "std", "count"]).reset_index()
summary_df["sem"] = summary_df["std"] / np.sqrt(summary_df["count"])
# ✅ Compute 95% Confidence Interval using t-distribution
summary_df["ci95"] = summary_df["sem"] * stats.t.ppf(0.975, df=summary_df["count"]-1)
return summary_df
# ✅ Compute statistics for Distance Correlation and Cosine Similarity
distance_summary = compute_summary(results_df, "Distance Correlation")
cosine_summary = compute_summary(results_df, "Cosine Similarity")
# ✅ Function to plot Mean ± 95% CI Error Bars
def plot_error_bars(summary_df, metric, y_label, y_lim=None):
"""Creates bar plots for each function type with 95% CI error bars."""
function_types = summary_df["Function Type"].unique()
for function in function_types:
plt.figure(figsize=(12, 7)) # Set figure size
subset = summary_df[summary_df["Function Type"] == function].sort_values(by="mean", ascending=False)
# ✅ Plot Mean with 95% CI error bars
plt.errorbar(subset["Model"], subset["mean"], yerr=subset["ci95"], fmt='o', color='red', ecolor='black', capsize=5, label="Mean ± 95% CI")
# ✅ Labels & Titles
plt.xlabel("Model", fontsize=18)
plt.ylabel(y_label, fontsize=18)
plt.title(f"{y_label} for {function.capitalize()} Functions", fontsize=18)
# ✅ Set y-limits if provided
if y_lim:
plt.ylim(y_lim)
# ✅ Rotate x-axis labels
plt.xticks(rotation=45, fontsize=18)
plt.yticks(fontsize=18)
# ✅ Display Mean ± CI values at the bottom
for i, txt in enumerate(subset["mean"]):
plt.text(i, min(subset["mean"]) - 0.02, f"{txt:.3f} ± {subset['ci95'].iloc[i]:.3f}", ha='center', fontsize=12)
plt.legend(fontsize=18)
plt.grid(axis="y", linestyle="--", alpha=0.7)
plt.tight_layout()
plt.show()
# ✅ Plot Distance Correlation (No specific y-limit)
plot_error_bars(distance_summary, "Distance Correlation", "Distance Correlation Mean")
# ✅ Plot Cosine Similarity (Set y-axis from 0.6 to 0.9)
plot_error_bars(cosine_summary, "Cosine Similarity", "Cosine Similarity Mean", y_lim=(0.6, 0.8))
In [40]:
import os
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
# ✅ Load the results file if it exists
RESULTS_FILE = "cluster_similarity_results.csv"
if not os.path.exists(RESULTS_FILE):
raise FileNotFoundError(f"❌ ERROR: The results file '{RESULTS_FILE}' was not found!")
results_df = pd.read_csv(RESULTS_FILE)
# ✅ Ensure that the dataset column is properly formatted
if "Dataset" in results_df.columns:
results_df["Function Type"] = results_df["Dataset"].str.extract(r'_(linear|exponential|quadratic|cubic)')
else:
raise ValueError("❌ ERROR: 'Dataset' column missing from results_df!")
# ✅ Sorting function
def sort_results(df, metric):
"""Sorts results within each function type based on the given metric in descending order."""
if metric not in df.columns:
raise ValueError(f"❌ ERROR: {metric} column is missing in results_df!")
return df.sort_values(by=[metric], ascending=False)
# ✅ Ensure sorting for both metrics
sorted_distance_df = sort_results(results_df, "Distance Correlation")
sorted_cosine_df = sort_results(results_df, "Cosine Similarity")
# ✅ Set visualization style
sns.set(style="whitegrid")
# 📊 **Boxplot for Distance Correlation**
plt.figure(figsize=(12, 6))
sns.boxplot(
x="Function Type",
y="Distance Correlation",
hue="Model",
data=sorted_distance_df,
palette="viridis",
dodge=True
)
plt.xlabel("Function Type", fontsize=18)
plt.ylabel("Distance Correlation", fontsize=18)
plt.title("Distance Correlation by Function Type and Model", fontsize=20)
plt.xticks(fontsize=18)
plt.yticks(fontsize=18)
plt.legend(title="Model", fontsize=16, title_fontsize=18, loc="upper left", bbox_to_anchor=(1, 1))
plt.show()
# 📊 **Boxplot for Cosine Similarity**
plt.figure(figsize=(12, 6))
sns.boxplot(
x="Function Type",
y="Cosine Similarity",
hue="Model",
data=sorted_cosine_df,
palette="coolwarm",
dodge=True
)
plt.xlabel("Function Type", fontsize=18)
plt.ylabel("Cosine Similarity", fontsize=18)
plt.title("Cosine Similarity by Function Type and Model", fontsize=20)
plt.xticks(fontsize=18)
plt.yticks(fontsize=18)
plt.legend(title="Model", fontsize=16, title_fontsize=18, loc="upper left", bbox_to_anchor=(1, 1))
plt.show()
# 📊 **Separate Boxplots for Cosine Similarity by Function Type**
function_types = sorted_cosine_df["Function Type"].unique()
for ftype in function_types:
plt.figure(figsize=(12, 6))
subset = sorted_cosine_df[sorted_cosine_df["Function Type"] == ftype]
sns.boxplot(
x="Model",
y="Cosine Similarity",
data=subset,
palette="coolwarm"
)
plt.xlabel("Model", fontsize=18)
plt.ylabel("Cosine Similarity", fontsize=18)
plt.title(f"Cosine Similarity for {ftype.capitalize()} Functions", fontsize=20)
plt.xticks(fontsize=18, rotation=45)
plt.yticks(fontsize=18)
plt.show()
/tmp/ipykernel_1505124/712002315.py:75: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(
/tmp/ipykernel_1505124/712002315.py:75: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(
/tmp/ipykernel_1505124/712002315.py:75: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(
/tmp/ipykernel_1505124/712002315.py:75: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(
In [44]:
In [41]:
import pandas as pd
pd.set_option("display.max_columns", None) # Show all columns
pd.set_option("display.max_rows", None) # Show all rows
pd.set_option("display.width", 1000) # Increase width
pd.set_option("display.colheader_justify", "center") # Center align headers
pd.set_option("display.float_format", "{:.4f}".format) # Limit decimal places
import os
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import h5py
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE
from sklearn.cluster import KMeans
from scipy.spatial.distance import pdist, squareform
from scipy.stats import spearmanr
from sklearn.metrics.pairwise import cosine_similarity
# ✅ Define dataset-embedding directories (Llama only)
input_dir = "/home/abbasali/synthetic data/generated_data_llama"
output_dir = "/home/abbasali/synthetic data/embeddings_llama"
# ✅ Step 1: Load original datasets (Generated Data Llama)
dataframes = {}
csv_files = [f for f in os.listdir(input_dir) if f.endswith(".csv")]
for file in csv_files:
dataset_name = os.path.splitext(file)[0]
df = pd.read_csv(os.path.join(input_dir, file))
# ✅ Drop non-numeric columns and ensure float32 type
df_numeric = df.select_dtypes(include=[np.number])
dataframes[dataset_name] = df_numeric.to_numpy(dtype=np.float32)
print(f"✅ Loaded {len(dataframes)} original datasets from {input_dir}.")
# ✅ Step 2: Load embeddings (Only Llama)
embedding_files = [f for f in os.listdir(output_dir) if f.endswith(".h5")]
embeddings_data = {}
for file in embedding_files:
dataset_name = file.replace("embeddings_", "").replace(".h5", "")
if dataset_name in dataframes:
with h5py.File(os.path.join(output_dir, file), 'r') as h5f:
embeddings = np.array(h5f['embeddings'][:], dtype=np.float32)
# ✅ Reduce embeddings with PCA to match original feature size
original_dim = dataframes[dataset_name].shape[1]
if embeddings.shape[1] > original_dim:
print(f"⚠️ Reducing embeddings for {dataset_name} from {embeddings.shape[1]} to {original_dim} dimensions.")
pca = PCA(n_components=original_dim)
embeddings = pca.fit_transform(embeddings)
embeddings_data[dataset_name] = embeddings
print(f"✅ Loaded {len(embeddings_data)} Llama embedding sets from {output_dir}.")
# ✅ Step 3: Compute Distance Correlation & Cosine Similarity
def compute_similarity(original_data, embedding_data):
"""Computes Distance Correlation and Cosine Similarity between original and embedding data."""
# ✅ Use the first 300 samples for consistency
sample_size = min(300, original_data.shape[0], embedding_data.shape[0])
original_sample = original_data[:sample_size]
embedding_sample = embedding_data[:sample_size]
# ✅ Compute pairwise distances
dist_orig = squareform(pdist(original_sample, metric='euclidean'))
dist_emb = squareform(pdist(embedding_sample, metric='euclidean'))
# ✅ Ensure both matrices have the same shape
min_size = min(dist_orig.shape[0], dist_emb.shape[0])
dist_orig, dist_emb = dist_orig[:min_size, :min_size], dist_emb[:min_size, :min_size]
# ✅ Compute metrics
correlation, _ = spearmanr(dist_orig.flatten(), dist_emb.flatten())
cosine_sim = cosine_similarity(dist_orig.flatten().reshape(1, -1), dist_emb.flatten().reshape(1, -1))[0][0]
return round(correlation, 3), round(cosine_sim, 3)
# ✅ Step 4: Compute Similarity for All Datasets & Sort by Cosine Similarity
similarity_results = []
for dataset_name, original_data in dataframes.items():
if dataset_name not in embeddings_data:
continue # Skip if no embedding available
embedding_data = embeddings_data[dataset_name]
distance_corr, cosine_sim = compute_similarity(original_data, embedding_data)
similarity_results.append((dataset_name, distance_corr, cosine_sim))
# ✅ Sort datasets by Cosine Similarity in descending order
sorted_datasets = sorted(similarity_results, key=lambda x: x[2], reverse=True) # Sort by cosine similarity
# ✅ Step 5: Reduce dimensions and generate **Side-by-Side t-SNE plots with Clusters**
def plot_tsne_side_by_side(sorted_datasets, original_data_dict, embedding_data_dict, title_prefix):
"""Reduces dimensions using PCA & t-SNE, then plots **side-by-side** t-SNE visualizations with KMeans Clusters."""
for dataset_name, distance_corr, cosine_sim in sorted_datasets:
original_data = original_data_dict[dataset_name]
embedding_data = embedding_data_dict[dataset_name]
# ✅ Use only the first 300 samples
sample_size = min(300, original_data.shape[0], embedding_data.shape[0])
original_sample = original_data[:sample_size]
embedding_sample = embedding_data[:sample_size]
# ✅ Apply PCA (reduce to 50 dimensions, if applicable)
pca_components = min(original_sample.shape[1], 50)
pca = PCA(n_components=pca_components)
pca_original = pca.fit_transform(original_sample)
pca_embedding = pca.transform(embedding_sample)
# ✅ Apply t-SNE (reduce to 2 dimensions)
tsne = TSNE(n_components=2, perplexity=30, random_state=42)
tsne_original = tsne.fit_transform(pca_original)
tsne_embedding = tsne.fit_transform(pca_embedding)
# ✅ Perform KMeans Clustering on t-SNE data
optimal_clusters = min(10, sample_size // 10) # Set a max of 10 clusters
kmeans_original = KMeans(n_clusters=optimal_clusters, random_state=42, n_init=10).fit(tsne_original)
kmeans_embedding = KMeans(n_clusters=optimal_clusters, random_state=42, n_init=10).fit(tsne_embedding)
# ✅ Create side-by-side scatter plots
fig, axes = plt.subplots(1, 2, figsize=(16, 8), sharex=True, sharey=True)
# ✅ Left Plot: Original Data
sns.scatterplot(ax=axes[0], x=tsne_original[:, 0], y=tsne_original[:, 1], hue=kmeans_original.labels_, palette="viridis", s=80, alpha=0.8)
axes[0].scatter(kmeans_original.cluster_centers_[:, 0], kmeans_original.cluster_centers_[:, 1], s=300, c='black', marker='X', label="Cluster Centers", edgecolors='white')
axes[0].set_title(f"Original Data - {dataset_name}", fontsize=18)
axes[0].legend(title="Cluster", loc="upper right")
axes[0].grid(True)
# ✅ Right Plot: Embedding Data
sns.scatterplot(ax=axes[1], x=tsne_embedding[:, 0], y=tsne_embedding[:, 1], hue=kmeans_embedding.labels_, palette="coolwarm", s=80, alpha=0.8)
axes[1].scatter(kmeans_embedding.cluster_centers_[:, 0], kmeans_embedding.cluster_centers_[:, 1], s=300, c='black', marker='X', label="Cluster Centers", edgecolors='white')
axes[1].set_title(f"Embeddings - {dataset_name} (Corr: {distance_corr}, CosSim: {cosine_sim})", fontsize=18)
axes[1].legend(title="Cluster", loc="upper right")
axes[1].grid(True)
# ✅ Final Formatting
plt.suptitle(f"{title_prefix} {dataset_name}", fontsize=16)
plt.tight_layout()
plt.show()
# ✅ Generate side-by-side t-SNE plots with clusters in order of highest cosine similarity first
plot_tsne_side_by_side(sorted_datasets, dataframes, embeddings_data, "t-SNE with Clusters - Llama Data")
import matplotlib.pyplot as plt
import seaborn as sns
def plot_tsne_one_by_one(sorted_datasets, dataframes, embeddings_data, title_prefix):
"""Generates separate t-SNE scatter plots for each dataset with clusters, one by one."""
for dataset_name in sorted_datasets:
if dataset_name not in dataframes or dataset_name not in embeddings_data:
continue # Skip if dataset is missing
df = dataframes[dataset_name]
embeddings = embeddings_data[dataset_name]
# Extract t-SNE transformed data
tsne_original = df["tsne_original"]
tsne_embedding = embeddings["tsne_embedding"]
# KMeans Clusters
kmeans_original = df["kmeans_original"]
kmeans_embedding = embeddings["kmeans_embedding"]
# Correlation Metrics
distance_corr = df["distance_correlation"]
cosine_sim = df["cosine_similarity"]
# ✅ Plot Original Data
plt.figure(figsize=(10, 6))
sns.scatterplot(x=tsne_original[:, 0], y=tsne_original[:, 1], hue=kmeans_original.labels_,
palette="viridis", s=80, alpha=0.8)
plt.scatter(kmeans_original.cluster_centers_[:, 0], kmeans_original.cluster_centers_[:, 1],
s=300, c='black', marker='X', label="Cluster Centers", edgecolors='white')
plt.title(f"Original Data - {dataset_name}", fontsize=18)
plt.legend(title="Cluster", loc="upper right")
plt.grid(True)
plt.show()
# ✅ Plot Embedding Data
plt.figure(figsize=(10, 6))
sns.scatterplot(x=tsne_embedding[:, 0], y=tsne_embedding[:, 1], hue=kmeans_embedding.labels_,
palette="coolwarm", s=80, alpha=0.8)
plt.scatter(kmeans_embedding.cluster_centers_[:, 0], kmeans_embedding.cluster_centers_[:, 1],
s=300, c='black', marker='X', label="Cluster Centers", edgecolors='white')
plt.title(f"Embeddings - {dataset_name} (Corr: {distance_corr}, CosSim: {cosine_sim})", fontsize=18)
plt.legend(title="Cluster", loc="upper right")
plt.grid(True)
plt.show()
# ✅ Generate separate t-SNE plots for each dataset
plot_tsne_one_by_one(sorted_datasets, dataframes, embeddings_data, "t-SNE with Clusters - Llama Data")
✅ Loaded 79 original datasets from /home/abbasali/synthetic data/generated_data_llama. ⚠️ Reducing embeddings for data_15_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_6_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_42_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_71_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_91_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_25_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_38_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_34_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_17_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_70_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_27_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_35_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_33_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_85_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_79_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_80_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_53_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_24_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_66_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_40_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_41_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_58_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_65_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_90_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_12_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_32_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_97_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_9_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_89_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_76_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_81_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_50_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_87_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_44_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_78_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_22_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_55_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_19_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_10_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_11_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_84_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_64_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_39_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_4_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_43_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_77_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_13_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_94_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_96_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_54_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_49_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_59_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_74_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_7_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_52_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_93_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_14_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_20_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_3_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_62_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_48_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_63_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_99_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_37_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_45_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_72_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_30_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_1_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_31_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_51_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_47_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_18_quadratic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_56_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_36_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_69_linear from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_75_exponential from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_98_cubic from 32000 to 3 dimensions. ⚠️ Reducing embeddings for data_83_exponential from 32000 to 3 dimensions. ✅ Loaded 78 Llama embedding sets from /home/abbasali/synthetic data/embeddings_llama.
In [78]:
import pandas as pd
pd.set_option("display.max_columns", None) # Show all columns
pd.set_option("display.max_rows", None) # Show all rows
pd.set_option("display.width", 1000) # Increase width
pd.set_option("display.colheader_justify", "center") # Center align headers
pd.set_option("display.float_format", "{:.4f}".format) # Limit decimal places
import os
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import h5py
from mpl_toolkits.mplot3d import Axes3D
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE
from sklearn.cluster import KMeans
from scipy.spatial.distance import pdist, squareform
from scipy.stats import spearmanr
from sklearn.metrics.pairwise import cosine_similarity
# ✅ Define dataset-embedding directories
input_dir = "/home/abbasali/synthetic data/generated_data"
output_dir = "/home/abbasali/synthetic data/embeddings"
# ✅ Load original datasets
dataframes = {}
csv_files = [f for f in os.listdir(input_dir) if f.endswith(".csv")]
for file in csv_files:
dataset_name = os.path.splitext(file)[0]
df = pd.read_csv(os.path.join(input_dir, file))
df_numeric = df.select_dtypes(include=[np.number]) # Drop non-numeric columns
dataframes[dataset_name] = df_numeric.to_numpy(dtype=np.float32)
print(f"✅ Loaded {len(dataframes)} original datasets from {input_dir}.")
# ✅ Load embeddings (Ernie, Bert, Gatortron, Roberta, T5)
embedding_files = [f for f in os.listdir(output_dir) if f.endswith(".npy")]
embeddings_data = {}
for file in embedding_files:
parts = file.replace("embeddings_", "").replace(".npy", "").split("_")
model_name = parts[0] # Extract model name
dataset_name = "_".join(parts[1:]) # Extract dataset name
if dataset_name in dataframes:
embeddings = np.load(os.path.join(output_dir, file)).astype(np.float32)
embeddings = np.atleast_2d(embeddings) # Ensure 2D format
if embeddings.size == 0 or embeddings.shape[0] < 2:
continue # Skip empty embeddings
# ✅ Reduce embeddings with PCA to match original feature size
original_dim = dataframes[dataset_name].shape[1]
if embeddings.shape[1] > original_dim:
pca = PCA(n_components=original_dim)
embeddings = pca.fit_transform(embeddings)
if dataset_name not in embeddings_data:
embeddings_data[dataset_name] = {}
embeddings_data[dataset_name][model_name] = embeddings
print(f"✅ Loaded {sum(len(v) for v in embeddings_data.values())} embedding sets from {output_dir}.")
# ✅ Compute Distance Correlation & Cosine Similarity
def compute_similarity(original_data, embedding_data):
"""Computes Distance Correlation and Cosine Similarity."""
original_data = np.atleast_2d(original_data)
embedding_data = np.atleast_2d(embedding_data)
if original_data.size == 0 or embedding_data.size == 0:
return np.nan, np.nan
sample_size = min(300, original_data.shape[0], embedding_data.shape[0])
original_sample = original_data[:sample_size]
embedding_sample = embedding_data[:sample_size]
if original_sample.shape[1] != embedding_sample.shape[1]:
return np.nan, np.nan
dist_orig = squareform(pdist(original_sample, metric='euclidean'))
dist_emb = squareform(pdist(embedding_sample, metric='euclidean'))
min_size = min(dist_orig.shape[0], dist_emb.shape[0])
dist_orig, dist_emb = dist_orig[:min_size, :min_size], dist_emb[:min_size, :min_size]
correlation, _ = spearmanr(dist_orig.flatten(), dist_emb.flatten())
cosine_sim = cosine_similarity(dist_orig.flatten().reshape(1, -1), dist_emb.flatten().reshape(1, -1))[0][0]
return round(correlation, 3), round(cosine_sim, 3)
# ✅ Compute Similarity for All Datasets & Sort by Cosine Similarity
similarity_results = []
for dataset_name, original_data in dataframes.items():
if dataset_name not in embeddings_data:
continue
for model_name, embedding_data in embeddings_data[dataset_name].items():
distance_corr, cosine_sim = compute_similarity(original_data, embedding_data)
similarity_results.append((dataset_name, model_name, distance_corr, cosine_sim))
similarity_df = pd.DataFrame(similarity_results, columns=["Dataset", "Model", "Distance Correlation", "Cosine Similarity"])
similarity_df = similarity_df.dropna() # Remove NaN values before sorting
# ✅ Find best and worst cosine similarity datasets per model
best_worst_per_model = {}
for model in similarity_df["Model"].unique():
model_df = similarity_df[similarity_df["Model"] == model].dropna(subset=["Cosine Similarity"])
if model_df.empty:
continue
best = model_df.loc[model_df["Cosine Similarity"].idxmax()]
worst = model_df.loc[model_df["Cosine Similarity"].idxmin()]
best_worst_per_model[model] = [best, worst]
# ✅ Visualization: 3D Scatter + t-SNE with KMeans Overlay
def plot_3d_tsne_side_by_side(best_worst_per_model, original_data_dict, embedding_data_dict):
"""Plots **3D scatter + side-by-side** t-SNE with **KMeans Clustering Overlay**."""
for model_name, (best, worst) in best_worst_per_model.items():
for dataset in [best, worst]:
dataset_name, _, distance_corr, cosine_sim = dataset
original_data = original_data_dict[dataset_name]
embedding_data = embedding_data_dict[dataset_name][model_name]
original_data = np.atleast_2d(original_data)
embedding_data = np.atleast_2d(embedding_data)
sample_size = min(300, original_data.shape[0], embedding_data.shape[0])
original_sample = original_data[:sample_size]
embedding_sample = embedding_data[:sample_size]
# ✅ Apply PCA
pca = PCA(n_components=50) if original_sample.shape[1] > 50 else PCA(n_components=original_sample.shape[1])
pca_original = pca.fit_transform(original_sample)
pca_embedding = pca.transform(embedding_sample)
# ✅ Apply t-SNE
tsne = TSNE(n_components=2, perplexity=30, random_state=42)
tsne_original = tsne.fit_transform(pca_original)
tsne_embedding = tsne.fit_transform(pca_embedding)
# ✅ Perform KMeans Clustering on t-SNE data
optimal_clusters = min(10, sample_size // 10) # Max of 10 clusters
kmeans_original = KMeans(n_clusters=optimal_clusters, random_state=42, n_init=10).fit(tsne_original)
kmeans_embedding = KMeans(n_clusters=optimal_clusters, random_state=42, n_init=10).fit(tsne_embedding)
# ✅ Left Plot: Original Data with Clusters
plt.figure(figsize=(10, 5))
sns.scatterplot(x=tsne_original[:, 0], y=tsne_original[:, 1], hue=kmeans_original.labels_, palette="tab10", s=80, alpha=0.8)
plt.title(f"t-SNE Original - {dataset_name}")
# ✅ Right Plot: Embedding Data with Clusters
plt.figure(figsize=(10, 5))
sns.scatterplot(x=tsne_embedding[:, 0], y=tsne_embedding[:, 1], hue=kmeans_embedding.labels_, palette="tab10", s=80, alpha=0.8)
plt.title(f"{model_name} Embeddings - {dataset_name} (CosSim: {cosine_sim})")
plt.show()
# ✅ 3D Scatter Plot of Original Data
fig = plt.figure(figsize=(10, 7))
ax = fig.add_subplot(111, projection='3d')
ax.scatter(original_sample[:, 0], original_sample[:, 1], original_sample[:, 2], alpha=0.6)
ax.set_title(f"3D Scatter of {dataset_name}")
plt.show()
plot_3d_tsne_side_by_side(best_worst_per_model, dataframes, embeddings_data)
✅ Loaded 100 original datasets from /home/abbasali/synthetic data/generated_data. ✅ Loaded 500 embedding sets from /home/abbasali/synthetic data/embeddings.
In [33]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
# Sample DataFrame (Replace this with your actual dataset)
# results_df = pd.read_csv("your_file.csv")
# Extract the function type from the Dataset column
results_df["Function Type"] = results_df["Dataset"].str.extract(r'_(linear|exponential|quadratic|cubic)')
# Set visualization style
sns.set(style="whitegrid")
# **Boxplot for Distance Correlation Grouped by Function Type**
plt.figure(figsize=(12, 6))
sns.boxplot(x="Function Type", y="Distance Correlation", data=results_df, palette="viridis")
plt.xlabel("Function Type", fontsize=16)
plt.ylabel("Distance Correlation", fontsize=16)
plt.title("Distance Correlation by Function Type", fontsize=18)
plt.xticks(fontsize=14)
plt.yticks(fontsize=14)
plt.show()
# **Boxplot for Cosine Similarity Grouped by Function Type**
plt.figure(figsize=(12, 6))
sns.boxplot(x="Function Type", y="Cosine Similarity", data=results_df, palette="coolwarm")
plt.xlabel("Function Type", fontsize=16)
plt.ylabel("Cosine Similarity", fontsize=16)
plt.title("Cosine Similarity by Function Type", fontsize=18)
plt.xticks(fontsize=14)
plt.yticks(fontsize=14)
plt.show()
/tmp/ipykernel_973/3504825753.py:16: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(x="Function Type", y="Distance Correlation", data=results_df, palette="viridis")
/tmp/ipykernel_973/3504825753.py:27: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. sns.boxplot(x="Function Type", y="Cosine Similarity", data=results_df, palette="coolwarm")
In [29]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
# Load and clean CSV
summary_df = pd.read_csv('summary.csv', header=1)
summary_df.columns = ['Model', 'Distance Correlation Mean', 'Distance Correlation Std',
'Cosine Similarity Mean', 'Cosine Similarity Std']
# Convert columns to numeric
numeric_cols = summary_df.columns[1:]
summary_df[numeric_cols] = summary_df[numeric_cols].apply(pd.to_numeric, errors='coerce')
# Set visualization style
sns.set(style="whitegrid")
# **Sort Data by Distance Correlation Mean in Descending Order**
sorted_distance_df = summary_df.sort_values(by="Distance Correlation Mean", ascending=False)
plt.figure(figsize=(12, 6))
# Scatter plot for means
plt.scatter(sorted_distance_df["Model"], sorted_distance_df["Distance Correlation Mean"],
color='red', s=120, edgecolors='black', label="Mean")
# Error bars
plt.errorbar(sorted_distance_df["Model"], sorted_distance_df["Distance Correlation Mean"],
yerr=sorted_distance_df["Distance Correlation Std"], fmt='none',
ecolor='black', capsize=5, capthick=1.5, elinewidth=1.5)
# Formatting
plt.xticks(rotation=45, fontsize=14)
plt.yticks(fontsize=14)
plt.xlabel("Model", fontsize=16)
plt.ylabel("Distance Correlation Mean", fontsize=16)
plt.title("Distance Correlation Across Models (Sorted)", fontsize=18)
plt.legend()
plt.grid(axis='y', linestyle='--', alpha=0.7)
plt.show()
# **Sort Data by Cosine Similarity Mean in Descending Order**
sorted_cosine_df = summary_df.sort_values(by="Cosine Similarity Mean", ascending=False)
plt.figure(figsize=(12, 6))
# Scatter plot for means
plt.scatter(sorted_cosine_df["Model"], sorted_cosine_df["Cosine Similarity Mean"],
color='blue', s=120, edgecolors='black', label="Mean")
# Error bars
plt.errorbar(sorted_cosine_df["Model"], sorted_cosine_df["Cosine Similarity Mean"],
yerr=sorted_cosine_df["Cosine Similarity Std"], fmt='none',
ecolor='black', capsize=5, capthick=1.5, elinewidth=1.5)
# Formatting
plt.xticks(rotation=45, fontsize=14)
plt.yticks(fontsize=14)
plt.xlabel("Model", fontsize=16)
plt.ylabel("Cosine Similarity Mean", fontsize=16)
plt.title("Cosine Similarity Across Models (Sorted)", fontsize=18)
plt.legend()
plt.grid(axis='y', linestyle='--', alpha=0.7)
plt.show()
In [26]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
# Load and clean CSV
summary_df = pd.read_csv('summary.csv', header=1)
summary_df.columns = ['Model', 'Distance Correlation Mean', 'Distance Correlation Std',
'Cosine Similarity Mean', 'Cosine Similarity Std']
# Convert columns to numeric
numeric_cols = summary_df.columns[1:]
summary_df[numeric_cols] = summary_df[numeric_cols].apply(pd.to_numeric, errors='coerce')
# Set visualization style
sns.set(style="whitegrid")
# **Improved Plot: Bar plot with Matplotlib error bars for Distance Correlation**
plt.figure(figsize=(12, 6))
ax = sns.barplot(x="Model", y="Distance Correlation Mean", data=summary_df, palette="viridis", ci=None)
# Add error bars manually using Matplotlib
for i, bar in enumerate(ax.patches):
x = bar.get_x() + bar.get_width() / 2
y = bar.get_height()
err = summary_df["Distance Correlation Std"].iloc[i]
plt.errorbar(x, y, yerr=err, fmt='none', ecolor='black', capsize=5, capthick=1.5, elinewidth=1.5)
plt.xticks(rotation=45, fontsize=14)
plt.yticks(fontsize=14)
plt.xlabel("Model", fontsize=16)
plt.ylabel("Distance Correlation Mean", fontsize=16)
plt.title("Distance Correlation Across Models", fontsize=18)
plt.grid(axis='y', linestyle='--', alpha=0.7)
plt.show()
# **Improved Plot: Bar plot with Matplotlib error bars for Cosine Similarity**
plt.figure(figsize=(12, 6))
ax = sns.barplot(x="Model", y="Cosine Similarity Mean", data=summary_df, palette="coolwarm", ci=None)
# Add error bars manually using Matplotlib
for i, bar in enumerate(ax.patches):
x = bar.get_x() + bar.get_width() / 2
y = bar.get_height()
err = summary_df["Cosine Similarity Std"].iloc[i]
plt.errorbar(x, y, yerr=err, fmt='none', ecolor='black', capsize=5, capthick=1.5, elinewidth=1.5)
plt.xticks(rotation=45, fontsize=14)
plt.yticks(fontsize=14)
plt.xlabel("Model", fontsize=16)
plt.ylabel("Cosine Similarity Mean", fontsize=16)
plt.title("Cosine Similarity Across Models", fontsize=18)
plt.grid(axis='y', linestyle='--', alpha=0.7)
plt.show()
/tmp/ipykernel_973/3071221424.py:19: FutureWarning: The `ci` parameter is deprecated. Use `errorbar=None` for the same effect. ax = sns.barplot(x="Model", y="Distance Correlation Mean", data=summary_df, palette="viridis", ci=None) /tmp/ipykernel_973/3071221424.py:19: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. ax = sns.barplot(x="Model", y="Distance Correlation Mean", data=summary_df, palette="viridis", ci=None)
/tmp/ipykernel_973/3071221424.py:38: FutureWarning: The `ci` parameter is deprecated. Use `errorbar=None` for the same effect. ax = sns.barplot(x="Model", y="Cosine Similarity Mean", data=summary_df, palette="coolwarm", ci=None) /tmp/ipykernel_973/3071221424.py:38: FutureWarning: Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect. ax = sns.barplot(x="Model", y="Cosine Similarity Mean", data=summary_df, palette="coolwarm", ci=None)
In [18]:
<class 'pandas.core.frame.DataFrame'> RangeIndex: 6 entries, 0 to 5 Data columns (total 5 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 Unnamed: 0 6 non-null object 1 mean 6 non-null float64 2 std 6 non-null float64 3 mean.1 6 non-null float64 4 std.1 6 non-null float64 dtypes: float64(4), object(1) memory usage: 368.0+ bytes Unnamed: 0 mean std mean.1 std.1 0 llama 0.515667 0.177299 0.876111 0.047025 1 bert 0.511960 0.170502 0.851030 0.170502 2 ernie 0.489020 0.187727 0.823740 0.076882 3 gatortron 0.498280 0.180313 0.861190 0.056885 4 roberta 0.503750 0.193689 0.832420 0.082608
In [ ]:
#to write file to html
import nbformat
from nbconvert import HTMLExporter
# Load your notebook
with open("synthetic_llama_plotting.ipynb") as f:
nb = nbformat.read(f, as_version=4)
# Convert to HTML
exporter = HTMLExporter()
body, _ = exporter.from_notebook_node(nb)
# Write the HTML to a file
with open("uci_machine_learning_llm_analysis_using_t5.html", "w") as f:
f.write(body)